I think selling is the hardest part of investing. Buying is, of course, critical to generating strong investment performance. Figuring out what to buy and when to buy it is what most people think of when they think of investing. But your returns will have as much to do with selling as buying. And buying is a fairly rational decision. Selling tends to be emotional. And that is why selling is the hardest part of investing.

In venture capital, thankfully, VCs don’t drive a lot of the sell decisions. I wrote about that back in 2009. Most sell decisions that really matter in a venture portfolio will be made by the founders and management of the portfolio company, including the timing of the public offering if that is where a company is headed.

But even so, I have struggled with the sell decisions, both personally and professionally, over the course of my career. I have held on way too long and watched a publicly traded stock literally go all the way to zero without selling it (ouch). And I have made the even worse decision of selling too soon and watching a stock go up three, four, five times from where I sold it.

So where I have landed on selling is to make it formulaic.

If we (USV) have to make a sell decision, we like to have a policy and stick to  it. We like to distribute our public positions as soon as we can, for example. That’s a policy and we stick to it. If you look at the SEC forms we have filed as a firm over the years, you can see that is what we do. It is a formula. It doesn’t mean that it is the right decision in each instance, but it does mean that, if we stick to it, we will do the same thing every time and the law of averages will work things out. We also let the people we work with know that is our policy so they are not surprised by it. What is worse is to make each decision emotionally and get them all or most of them wrong.

Personally, I like to dollar cost average out of a stock (sell a position over time instead of all at once) and I also like to hold onto some of the position for the very long term (schmuck insurance). I have a formula for the disposition of public stocks I get via distribution from USV and the other VC funds we are invested in. We execute the formula time and time again. It takes the emotion out of the decision and it works better for us.

The thing I have learned about selling is that it is almost impossible to optimize the sell point. You need a crystal ball and you need to know something that others don’t know. That is either impossible or criminal. So I don’t try to optimize it. I try to make it formulaic and systematic. It works better for me and I think it may work better for you too.

#stocks#VC & Technology

Comments (Archived):

  1. Brandon Burns

    Someone woke up thinking about Twitter this morning…

    1. fredwilson

      No. I would be buying if I didn’t already have a very large position!

      1. Mitch Lauren

        You can still buy as it is going to be affordable for everyone to buy. Forget the value now, they are working on calculating the cost of all the machines, minus the depreciation costs) and coming up with a fixed price. That is how bad it has become. Popular still but no real value.

      2. Hu Man


      3. LE

        I’ve come to the (obvious) conclusion, with the stock in question, that one of several things is happening. a) There is something you know that we don’t. b) You are acting highly emotionally not just emotionally. c) You are staying to the bitter end to be consistent with your previous thoughts on the stock [1] d) I don’t know what you mean by “large position” particularly what percentage of your assets it is (ie 5%, 25%, 50%). e) You needed to respond to Brandon to not send the wrong message to others f) You need an intervention.With respect to “f” I’ve thought of this before multiple times. But then again I don’t know the answer to “d” and “a”. Don’t know who the schmuck is in this case.I don’t expect you to answer this but I wonder how much gotham girl agrees with how you make sell decisions (since 50% of the stock is hers).I think I am going to go out and buy some of that stock today because I’ve decided that the upside appears greater than the downside risk.[1] This is similar to me wanting to justify a certain candidates behavior in face of overwhelming evidence to the contrary as well as ignoring downside risk.

        1. Mark Essel

          Hah, great comment. I picked up some shares, at $23.40 🙁

      4. JamesHRH

        No kidding. Twtr is one person away from being a monster.Sadly, that person excels at self-preservation.

        1. Salt Shaker

          TWTR will be sold, Q is at what price. Too large a sub base and too much data not to have value. (Same as Pandora in the streaming space.) At this point I hope it continues to drop so I can buy more shares, say $14-15, to offset my current loss position when I ultimately do sell. I likely won’t ever breakeven but it’s part of a strat of managing downside risk.

        2. Twain Twain

          Do you mean monster as in a $100 billion company or monster as in a nightmare for investors and potential acquirers?* http://uk.businessinsider.c…Who’s the person excelling at self-preservation?

          1. JamesHRH

            Monster business.Jack

          2. Twain Twain

            Here’s Forbes advice for Jack to merge TWTR & SQ. He’s likened Twitter to a “town square” before so maybe that’s the end game?*…Really, no one knows what will happen with TWTR. Every few months, it gets a temporary share price boost on speculations of being acquired or securing some license to stream a handful of sports games and then those plans goes south.

          3. JamesHRH

            The BS cloud around twitter is immense.It is:- live, direct access to an audience- for public entities ( individuals or orgs )It is a fee for service distribution company, not a competitor to Medium or Snapchat or whatever currently fuels Ass-Jack’s latest fog machine efforts.Twitter is to the AP news wire ( and it should be funded by the people who need access to their audience on a fee for service basis, not ads )……what FB is to HS yearbooks ( which were funded by local ads ).End of story. Probably worth $50B over time.

          4. Twain Twain

            They need to source a Head of Product and they need to give that person a proper mandate.From Bberg: “Within his management team, he’s largely delegated command of Twitter’s strategy to chief financial officer Anthony Noto.”*

          5. JamesHRH

            Can’t do that w current CEO who promotes himself as Product Visionary.Don’t believe the delegation stuff. He can say that, but when the CFO doesn’t control the product, he doesn’t control the strategy.Classic political smokescreen from a master fog maker.In the 1990’s, the Calgary Flames had a hyper-political, self preservation oriented General Manager named Doug Risebrough. They churned through several coaches, a star player who was widely beloved, a star goalie (most important position in hockey), the team President and finally the team CEO. All this happend over the course of 6 or 8 years of utter turmoil, during which an extremely talented team underperformed wildly.They finally fired Risebrough because they had fired everybody else.Then the team improved.It appears that Jack is pulling a Risebrough on the Twitter BoD. The company is in total turmoil: mass executive defections roiling the technocrati universe, now takeover rumours bouncing the stock around in the financial universe, I am sure he will cook up two or three other red herrings and then, finally, he’ll be the only thing they haven’t changed.Then he’ll be gone and the company will have a chance to move forward.Political wizards put their self preservation ahead of all other interests: the company itself, the investors, the employees and the customers.I can’t stand any of them. Ass-Jack included.

          6. Twain Twain

            Bberg: “Brad Slingerlend, who manages the global technology fund for Janus Capital Group, holds Facebook, Google, Amazon and Apple–but sold Twitter after Dorsey took over. Social networking, he said, is a business that depends on a network effect to generate growth. If the network’s momentum starts to ebb, it’s difficult to recover.”The article credits Noto with Twitter’s “live” strategy and the NFL streaming deal:*

          7. JamesHRH

            Come on Twain, those are deals and features.Its like telling me that I am in charge of strategy at Ford if I brought in a couple of deals for delivery vans.

          8. Twain Twain

            Yeah, “C’mon, Twain!!!” :).I have no positions in Twitter so no vested interest in whether they do well, or not. In their first board call together, Noto did a better job selling Twitter to me. He sounded “on the money” about what Twitter is and does.Here’s what I found interesting recently. Jack positioned himself as a “Steve Jobs 2.0”. Meanwhile, Zuckerberg hasn’t let himself be compared as “Gates 2.0” (even though MS’ investment in FB was what took it to unicorn status before its IPO) but as himself.At Apple’s most conference, it was surprising they gave a slot to Instagram Stories to showcase the iPhone 7 camera rather than to Twitter Moments and Periscope.It’s unclear how Twitter video streaming over Apple TV, XBox and Amazon Fire will work out.The other interesting thing is Twitter cancelled its developer conference this year:*…As a business case study, there are lots of interesting things in Twitter’s arc.

          9. Mark Essel

            Thats not a good sign

          10. Mark Essel

            I’d like to see that.

      5. Chimpwithcans

        Live Sports a factor?

      6. Mark Essel

        Thats good to know. I’m rooting (and betting) on twitter.

  2. William Mougayar

    You’re never happy, selling.

    1. LE

      With boats, they say happiest day is when you buy and when you sell. I hate that expression I wasn’t happy the day I ‘sold’ my boat.Actually interesting about that particular boat. I had a boat once that I was offered $x for. (24 Foot Sea Ray). At the time $x was 1/3 of what I paid and perhaps 70% of what I wanted. I turned it down thinking it was worth more and decided to wait it out. Meanwhile I wasn’t using it and had to pay storage and it started to deteriorate. A few years later there were no buyers so I just ended up walking into the dealer (Marine Max iim) and gave them the title. Just walked away from it.Recently, this year, I bought a nice car and decided that I had had made a mistake. So I sold it to another dealer, took a loss obviously (less than a month later) and bought another new car that I really wanted. I then got the car manufacturer to cover my loss on the trade in. True story. The money wasn’t as nice as the victory.

      1. William Mougayar

        My saying applies to the stock market, especially when you trade options. Getting out is more important than maximizing your profit.

  3. WA

    Displine consistency. Bias resistant. Most def.

  4. falicon

    Agree – your formula approach is exactly how I handle (public) investing now as well.I did this with Bitcoin as well – bought a bunch at a set price, set a sell price at purchase time and when it hit that, I sold all but 1 (basically sold at a price that allowed me to make a small amount and still keep 1 full bitcoin as if I got it for free…and now that 1 remaining one is along for the long play/ride [schmuck insurance as you call it])…which also keeps me interested in it’s market (and waiting for another fall to buy more). 🙂

    1. LE

      which also keeps me interested in it’s marketYep many people (not me) do this with stocks as well. They buy a small amount just so that things are interesting. I guess it’s similar to some types of sports betting.

  5. VincentWright

    Why not bring in a “Sell Team”? (Educate them on the final selling objective you want and let them sell sans emotional attachment…)

    1. Adam Sher

      That would dilute the exit fee and carried interest. Equally true and less cynically, investment banking exists to solve this problem. It is usually too expensive to keep a sell team in house because one company does not create enough deal flow.

  6. Lawrence Brass

    Can a fund manager take a personal position at the same time of a fund portfolio sale?

    1. JamesHRH

      Yes – happens a lot.Jim Breyer made HHuuuuigggeee glue in FB, investing beside his fund.

    2. Salt Shaker

      Slippery slope. There are SEC restrictions and financial institutions may have additional policies and restrictions that are policed internally by compliance depts (e.g., restricted lists, minimum hold periods, conflict of interest).

  7. Tom Labus

    maybe the hardest part of investing. I’ve rode the same stock up and down three times without selling. Someone suggested hiring a guy with a 2 x 4 to remind me of that part of the equation.

    1. pointsnfigures

      Yesss, and sometimes the 2×4 doesn’t work.

    2. JamesHRH

      Absolutely the truth.

    3. Girish Mehta

      Getting out of a stock that is on its way down is difficult because of our preferred tendency for “loss aversion” (Prospect Theory). Selling a stock that is down formalizes/realizes a loss…while holding onto it keeps alive the hope that it will recover. Thats loss aversion right there. The problem is that the stock does not know that you love it.A stock that is down 90% is a stock that was down 80%…..and then got cut in half.A trait that is common to many successful traders is “Cut your losses, and let your winners run”. They keep very tight stop losses, and realize the loss immediately when the stop loss is hit. This is basically a system to fight against the tendency of loss aversion.Investing is different from trading, so you pick the point of getting out – both on the way up and way down – differently, but still useful to have a point of view on getting out before you get in.The Robert De Niro character in Ronin – https://uploads.disquscdn.c

      1. Tom Labus

        Hard to fall out of love but have gotten a little better

      2. Tom Labus

        Great movie too

        1. Girish Mehta

          Yes. Ronin and Heat – Two De Niro movies from the 90’s that are a great re-watch. John Frankenheimer and Michael Mann respectively.

  8. Gregg Freishtat

    I consider the decision of when to sell my formerly private stock that is now liquid and public a bit of a champagne problem…. The hard ones are when/if to sell verses raising more capital. While the same market timing issues are usually in play there is no formula for the former….

  9. vishal k gupta

    It would be great to understand your formulaic selling logic on a high level. Obviously theres no one formula that always works.Selling has always been the hardest part to me. I usually try to come up with some idea of what I think the company should trade for (based on known values/peers in the market), then sell when the stock price is far over what I think it’s actually worth.

  10. Mike Cautillo

    As a trader I can certainly validate this process. I have much of my exits and risk per-determined before entering any positions. This is quite effective (for me) because of my previous experiences- it becomes very difficult to make objective decisions under stress (from market action) due to the many different cognitive biases we draw on at that time. Crazy thing is, we’re mostly unaware of them at the time, hence why Freds’ approach is logical. Nice to see it being effective in the VC world also.

  11. JaredMermey

    What about selling stock to secondary buyers on the private market?

  12. Seth Godin

    In “Finite and Infinite Games”, James Carse writes about the difference between games we play to play, and games we play that we must win. Playing catch with your son is an infinite game, 1 on 1 basketball for money is a finite game.In my experience, those that seek to make a life of projects, of building things, of engaging and changing the world around them–we’re playing an infinite game.If you view this stock or this company as the one and only time you’re going to play, then selling to win (full maximization) is a logical plan, even though the odds are against you and the psychic energy is likely to be negative for a really long time.But if you’re playing to play, then selling appropriately is different than selling to win.Maybe I’m rationalizing, but over the years, this made my decision (thanks Fred) to sell Yoyodyne a lot more palatable.All the time we spend second-guessing ourselves is time we don’t spend guessing about the next thing.

    1. LE

      Interesting in a way this sounds like “make it up on volume”. That is make more deals don’t worry about maximizing each deal. If that is the case, I agree.

    2. PhilipSugar

      This is a great post as usual Seth. It is so correct. It is why it’s important to have at least a couple of people involved that have been through it before. The first time is so much harder than the third or fourth.

    3. JamesHRH

      Any chance you could repost in English?What do you think Bill Gross’ opinion is on this topic? He’s the ultimate play to play investor & I bet the Overture/Google situation eats his soul.

    4. JaredMermey

      I believe most LPs would disagree.

    5. Vivek Kumar

      Great perspective. You learn more and become better by playing an infinite game. So over the long term one might win more games by playing to play.

    6. Twain Twain

      And therein is why training Machine Intelligence is NOT like training the intelligence of kids. The first involves finite objectives. The latter is all about the infinite potential of both the parents and the kid.The AI is contained within set time limits and risk bounds.https://uploads.disquscdn.chttps://uploads.disquscdn.c

    7. Mark Essel

      Time is finite so many games are too, even some of the ones we believe are infinite.Playing only to win is a little silly after a certain life style threshold is achieved (comfortable living expenses covered indefinitely). “Behold I stand on a larger pile of money!” We can’t take it with us.It’s getting to that independence threshold that feels finite. Every sale is critical until you cross it.

  13. pointsnfigures

    When I saw the title of this post, I initially thought it was about the practice of “selling”-which is also very hard and underrated.What you are talking about is “getting out”. Getting out is an acquired skill. Sometimes I have gotten out of positions really well. Sometimes I have sucked at it. When I was actively trading in the pit, the key was always knowing how to cover your losers-where could you go with them?In VC, I totally agree with you on principles. Get out based on a policy and never go away from it. That makes things predictable. Since there is so much volatility with investments, and so much volatility with harvesting them it makes a lot of sense-especially to get the public out first which I agree with strongly.Once you personally own the stock, it’s actually harder. Instead of fighting for your investors, you are fighting for yourself and all the psychological baggage that comes with it. Mental accounting comes into play, and losers feel far worse than winners.

    1. LE

      If it’s your money sure there is often more of an emotional element going on.Otoh if you enjoy the game then just the same you don’t like to lose and like to be smart and do smart things. People play sports and games all the time where there is no money lost or gained but the pride of winning or the agony of defeat.There are also advantages of having your own money on the line vs. someone you are acting on behalf of. You don’t have any explaining to do and your reputation doesn’t take an impact if you bet wrong. Consequently you can actually (oddly enough) take on more risk. Because the downside doesn’t impact others. [1]I had someone recently (actually this has happened a few times) offer me more money if I achieved a certain goal). I told them they were wasting their money and didn’t need to do that. My motivation is not driven (in that case) by whether I make a bit more money or not. In fact I actually find that (once again in this particular case) counter productive since it sends the message that my interests are not aligned with theirs.[1] God knows I hate to do things on behalf of my mother. Have to be super careful much more careful than I would with my own money. And not because she needs the money or that the money is significant either.

      1. pointsnfigures

        When you get out, remember “Pigs get slaughtered”. Had a trade once. I was short. Went home big short overnight. Woke up in the morning and I was up 6 figures and thought I had the world licked. Lost 6 figures.

        1. LE

          It’s the hogs that get slaughtered the pigs get fat! I guess that’s what you meant.Here is another one. I don’t have the yiddish but it’s translated from the yiddish:”If you are going to eat like a pig, let it drip from your beard”.Would like to hear more trading war stories if you have them.

          1. pointsnfigures

            I should write a book…..most of them are terrifically hilarious. Like the time we bet thousands of dollars (high 5 figures) on if someone could throw a 16″ softball across the Chicago River. Yes, you can.

          2. LE

            You know I have a theory on gamblers and their streaks. I think there is truth to the fact that someone’s brain actually works better and smarter when they are in that high mental phase of winning. That is why when you are hot you are hot and when you are not you are not. This also works with sales and cold calling and I would imagine picking up women (don’t do that and never have but I think I’m on target). You are just sharper both mentally and physically when you are on a success streak. The positive emotions can float a great deal of boats. Negative emotions are excess weight.

          3. Adam Sher

            One of my old tennis coaches called what you described, “controlled aggression.” You need to get yourself worked up to the point where you are somewhat pissed off, have elevated your heat rate, and are looking to kill. BUT you are still in control of your game plan.Other players (across most sports) call this being in the zone. You also do not think much, and let your training work almost on autopilot. This is one of the biggest reasons interviews with athletes are so bad. “I tried not to think, and just execute,” is getting yourself in the zone. It sounds trite, and the statement massively undersells the experience of being in the zone, not thinking, and executing. It’s euphoric, ecstatic, et. al.David Foster Wallace wrote about this in, “How Tracy Austin Broke My Heart.”

          4. JamesHRH

            That’s for sure a sample of a great book.

          5. Twain Twain

            Yes, you should write a book!

  14. BulldogSpirit

    Love the concept of “shmuck insurance”, Fred. I’m going to use that one. The tricks the mind can play just by having a little investment left in a stock that goes to the moon…:-)

    1. LE

      You’d be surprised at how letting someone share in the upside of something they are selling is a great way to overcome an objection to selling. I have done this and don’t have a name for it. But I probably wouldn’t present it as “schmuck” insurance.

      1. Adam Sher

        Hope note, back end piece, or JV. It depends on the seller’s remaining ownership and priority of cash flow. This happens in commercial real estate a lot among smaller transactions (e.g. sponsor funds, operating partners, and discretionary funds < $500m equity). If the seller is confident in his/her valuation, the seller will sometimes sell the majority share at a lower price and stay in the deal as an LP; or the seller will receive some amount of cash flow above a preferred return to the new equity (e.g. back-end piece, share of the ups, or hope note*).*Hope notes are usually for lenders who acquiesce to the borrower buying back debt at a discount but not fully ready to admit “defeat.”

  15. Matt Kruza

    Good overall approach when you are talking about investing (as you mention). As you allude to not that helpful for sample size n =1 (most entrepreneurs.. especially in the moment of a big sale decision of a company they worked on for years if not a decade!)

  16. iggyfanlo

    As a founder selling post-IPO stock, I put in place a simple system (market price MINUS IPO price) *0.1% every week capped at SEC trading limits per time period for first year… took the gun out of my hand

  17. sigmaalgebra

    A candidate selling formula for selling public stocks:From experience and judgment, define a few categories of stocks. E.g., maybe Lyft, Amazon, Twitter, and some blockchain company may be in different categories.For each category, from stocks you own and others, get the stock price data starting at IPO. Scale the data to the same initial value, say, 10. At each point in time after the IPO, average the scaled values. Find the time of the maximum. Maybe that is 4 months, 3 days, 5 hours, 23 minutes, and 10 seconds. So, for a stock in that category in your portfolio that did an IPO, that’s your time to sell. Done.For something better, maybe look also at the options prices and do something similar.Back test this analysis to see if it holds going back several years. If the analysis does not do well on the back check, then junk it. Yes, be careful to avoid the sin of over fitting, likely not a problem with what I described.Then in the future, continually repeat the analysis to confirm that the selling time still looks good.

    1. Mark Essel

      I dig it.Not overly complex, but it looks to exploit the historic time flux of prices post IPO by class.I worry an outlier could drastically shift that window though (one big winner). Maybe median filter out the extremes.Is there a case where you would break this rule (an investment is doing very very well so only sell a portion of it)?

      1. sigmaalgebra

        I worry an outlier could drastically shift that window though (one big winner). Maybe median filter out the extremes.For each category of stocks and at each time, will have an average. So, can get the confidence interval on that average — traditional ways based on the central limit theorem or maybe some newer ways, distribution-free, based on versions of resampling..If for that category of stocks, the confidence intervals for all the times look nicely small, at least near the candidate selling time, then f’get about outliers, black swans, fat tails, median filtering, etc. and smile on the way to the bank. If the confidence intervals look awful, do a different analysis.I’m skeptical about median filtering: Now, maybe if we could also do the same filtering on the stock we wanted to sell! But, alas, we can’t until we are in the future and can look back.The data that looked like an outlier was part of the data, on the real world. To me that the data looked like an outlier tells me to build a better model, one the includes more of reality.My suggestion does not handle all conceivable real situations. Instead, the suggestion is hoping for some good luck: Maybe that simple suggestion will appear to work and also actually work; in that case smile on the way to the bank. E.g., give luck a chance!Besides, the problem Fred wanted to solve wasSelling tends to be emotional. And that is why selling is the hardest part of investing.and for a solution Fred wantedSo where I have landed on selling is to make it formulaic.If we (USV) have to make a sell decision, we like to have a policy and stick to it.It looks like Fred is trying to minimize his emotional strains as much as maximize his financial gains — I didn’t intend that to rhyme; at least there is no awful alliteration. So, right, he has a problem in multi-objectives where one of his objectives is to use the KISS method — Keep it Simple Stupid.there a case where you would break this rule (an investment is doing very very well so only sell a portion of it)?That would add to Fred’s emotional strains! So, first-cut, no.Selling a portion, etc., may be better considering some personal utility functions, say, when the first million made is the most important. But, maybe that is not a consideration for Fred. And, besides, again, Fred wants something to minimize emotional strains, something “formulaic”, something simple.For high end ways to make money with mathematical models in the markets, call up James Simons. Here I wasn’t trying to compete with Simons.Also, besides, with the usual assumptions about market prices, e.g., Brownian motion, the calculation of my suggestion would come out that it mostly didn’t matter when sell.So, to do the analysis, basically we are assuming that there is some simple pattern of a short-term, post-IPO pop.For more, Fred could try to get some estimates of variance and covariance between pairs of stocks and apply the Markowitz work.For more, mostly we are taking about stocks in the USV portfolio where, of course, USV was a venture investor, maybe back to the Series A, and maybe has long had a board seat. So, for that company, Fred is an insider. So, he should know more. Then maybe one reason Fred wants something “formulaic” is to have the fact that he is selling not hurt the value of the stock.Fred and USV are successful investors. Markowitz got a Nobel prize for his little application of optimization with a quadratic objective and linear constraints. Simons got rich. I just made a simple suggestion to have something “formulaic”!My approach to getting rich is my startup which is based on using some math to provide a better service to get ad revenue to build a valuable company. So, I’m not trying to make money in the markets of equities of companies run by other people.All that Brownian motion stuff that says that quite broadly can’t make money in markets doesn’t apply in any sense to what I can derive with math, program into Windows, etc.! Instead, in the sense of investing, I’m the ultimate insider — I’m the founder, CEO, CTO, CIO, Chief Programmer, Lead Software Engineer …, janitor and know everything there is to know about my company! And the value of my company can’t be hurt by the opinions of Wall Street analysts! Also, my company has no destructive internal politics, no goal subordination, no possibility of employees stealing intellectual property, etc.! And not least, no board of directors can fire me!

        1. Mark Essel

          “call up James Simons.” Hah, I applied to Renaissance back in the early 2000s, it’s right down the road from my house. One of my coworkers at PRA (small engineering firm) got an offer but turned it down. Shame on you Jimbo! He could have helped hire me.How goes the search for early customers for your company? I hit a wall getting early buy in, back in 2009 (twitter contextual ads), but that lead to other startups and now I’m hooked (take that career as an aerospace R&D engineer).My own failed quietly (cofounder), the first I joined as employee #5 (video engineer) got purchased (no pay window), and my current startup is a few days away from launch (CTO) – if you have access to an iphone and are in NYC, keep an eye out for Vigilante next week.

          1. sigmaalgebra

            I’ve been in outrageously bad, independent, random exogenous interruption mode and, AFAIK, am now out of it and able to get back to work.The work requires that I get the material back between my ears. My documentation is so good and nearly all the code so simple that’s doable, but, still, there’s some work involved.But, due to interruptions, too often I’ve gotten the stuff between my ears again only to be interrupted again and, thus, lose the effort. No fun.So, I confess, for a week or so, I’ve paid too much attention the POTUS election.Without that, or interruptions, I will get back to work effortlessly — just sit for a while, get bored, and then eagerly get back to work! Pushing myself would also work but really is not needed: If I can beat back the interruptions, then I WILL effortlessly do the work.The core chunk of code, I want to check one more time. So, I will go through the math and the code, put a lot of print statements in the code, and then from the output check the arithmetic by hand and/or with some other code. I’ve written such other code before but now just want to do it one more time.Then there are a few trivial things I want to change in the UI.Then I will load some more initial data into the database and do, or finish, a really good alpha test.Then on to beta test. At HN I quickly got a nice list of beta testers. Maybe I will also invite beta testers from elsewhere!Then get more initial data. I’ve collected some good sources and already have some quite good data but will need to stuff it into the database. For that, I want to inject the data directly into some SQL tables and bypass some calculations that will usually be used (already programmed, running, used, etc.).Then get a first on-line server. My development computer is not quite powerful enough to be such a server.The first server might have a good Asus motherboard, an 8 core AMD processor at 4.0 GHz, 32 GB of ECC main memory, several rotating hard disk drives, good cooling.I did some cooling algebra: Given power to be dissipated in W, what air flow rate in cubic feet per minute need for a given temperature change F in the cooling air. So, maybe to dissipate 1000 W with only 10 F increase in temperature of the cooling air, what air flow rate do I need? Then make sure the case can provide that air flow rate easily.I believe I will tweak my Web site log file, not keep using Microsoft’s solution but roll my own.The advantage of Microsoft’s solution is that it serializes, that is, handles the problem of conflicts among several other programs wanting to write at about the same time. The disadvantage of the Microsoft solution is that it writes a lot of gibberish for which I have no documentation. So far I’ve just used some simple software to rip out the gibberish, but for production I want something better.My solution will be just to run a little program that will receive a log file entry via TCP/IP sockets and write the entry to a file. For the serialization, just use the TCP/IP FIFO input queue.For my Web site session state store, I’ve have such software running now — the log server will be simpler, and I can start with the code of the session state store and throw away what I don’t need..As for my session state store, will implement some commands I can give to the log server, e.g., report usage statistics, close present log file and start another one, and shutdown. We’re talking an easy afternoon.Then contact some ad networks and see what I have to do to run ads and do that.Do the usual, routine things, e.g., static IP address, domain name, trademark name, “doing business as” with the county, business bank account, initial chat with a CPA to see what I will need in initial record keeping, contact a certification authority (CA), business e-mail account, etc.Maybe implement HTTPS although my UI is so simple I suspect that HTTP is okay for now.Then go live. Contact the beta testers and let them know it’s live. Get publicity, etc. I’ve done some good things on publicity already.It’s all trivial, and in total a small fraction of the work I’ve done so far.But I don’t want to go live and in, say, six weeks suddenly discover some problem that will take me a month and have to shut down for that. So, before going live, I want to measure twice, saw once.Except for the core math, which I invented and is easy for me and the core software for that math, really, scientific-engineering software where I started my career and is also easy for me, the project is just an especially simple Web site. Especially given my additional checking of that core software, there’s no good reason the site should have any serious problems.If the usage explodes, then to handle it I will implement login name and password, and as my capacity increases draw from a FIFO queue of people applying to be users. With plenty of capacity, I will drop the need for login but retain the option for an additional, possibly good, feature, that will need login, for users who want to use the feature.As far as I know, all the work so far is fine. My development computer had some problems, were some of the interruptions, but AFAIK I’ve got them all solved fully or well enough for now.One advantage is the math: I doubt that anyone else will duplicate or even roughly equal my work.The crucial core of this project is some original math I derived with some advanced prerequisites. The math is NOT computer science, and nearly no one in computer science, student, professor, or chaired professor, has the math prerequisites to understand my math research even if I tried to explain it to them.I know some pure math people who could understand what I’ve done, but no way would they be interested in entrepreneurship. Also, they don’t know the computing, relatively simple now even for commercial computing but, still, they don’t know it and very definitely wouldn’t want to take more than 10 minutes to learn it.If some mathematicians were to try to duplicate or equal my project, they’d have to hire a CIO and a software development team, and that would cost bucks for which they would want equity funding, but the most equity funding they could get would be $0.00.As I’ve explained far too often at AVC, a project like mine, an applied math project, needs some computing, and the computer scientists do understand the computing, but they don’t understand the math. With at most just very few exceptions, the computer science people, even the chaired profs, just don’t know much math.To people not deep into both math and computing, such remarks of mine are no more welcome than a bucket of cold water on a kitty cat.I’m a solo founder with a tiny burn rate. Any reasonable early revenue will make me okay life style profitable. For more, an average of one user a second 24 x 7 will get me into the mode of a new SUV and a new supercharged Corvette, opera tickets, a new house, a nice server farm, and nice lifestyle mode.I’ve learned a solid lesson: No one on Sand Hill Road will be the least interested in me until long after I could no longer be interested in them:If I am successful, Sand Hill Road will call me, and for nearly any such call I will be able to check my records and tell them when they either turned down my initial contact or just ignored it.If they can’t understand my project now, then no way do I ever want to report to them on a BoD — they have long since proven they can’t understand my project either now or later.The Sand people have a fundamental problem: They desperately need exceptional projects, but except for just luck they have no very good way to identify such a project even when it comes across their desk.Thanks for your interest.

          2. Mark Essel

            sounds great, really excited to hear how your early release goes.Also happy to beta test if you need more data. I wouldn’t worry too much about scaling it up until you have to (pretty well known approaches to scaling workers behind a load balancer, and also your db should get you very very far just vertically – aka bigger systems).”One advantage is the math: I doubt that anyone else will duplicate or even roughly equal my work.”Patent or Trade Secret?

          3. sigmaalgebra

            Trade secrets. Just don’t tell anyone. If there is theft, then can still drag in the lawyers.The software and server farm architecture were designed to bequite scalable from the beginning.Yes, I’ve had in mind for later a load balancing front end IP router. Last time I did the Internet data rate arithmetic, 10 GbE would take me a very long way. But now we can get, what, 40 GbE, 100 GbE or some such? And there is much more that can be done with multiple front end IP routers for capacity.In the architecture, there is no need for session affinity. So, in a session for a user, the load balancing can send the packets from the user’s next HTTP GET or PUT to any Web server. Yup, I’ve programmed a good Web session state store — dirt simple (just two collection classes and TCP/IP sockets with object instance de/serialization), blindingly fast — simple arithmetic indicates that one rack of servers should be enough to serve the world. So, right, I didn’t use Redis — DIY instead. Simpler. Easier. Faster.The internal servers can all get a lot of reliability and performance from just simple sharding.There can be some significant hardware parallelism even within the work for a single request from a user — a working, relatively simple, first-cut at that is programmed and running now.There is one SQL table that could be replaced, or just replicated, in production as just a direct access file, write once a week and read thousands of times a second, and take fantastic advantage of solid state disks (SSD) — now up to 15 TB each! My back of the envelope arithmetic indicates that a rack of 20 of those SSDs could serve the world — right, ballpark $500 billion capitalization style.In production, the busy SQL tables are write only and, thus, can be parallelized nearly without limit.You are correct: For SQL reads, SQL Server’s use of main memory to cache tables stands to do wonders for performance, say, with some server with ballpark 1 TB of main memory, maybe 24 cores, etc.It’s scalable!Thanks for your interest. You are now on the beta test list!

  18. kevando

    > I have held on way too long and watched a publicly traded stock literally go all the way to zero without selling it (ouch). And I have made the even worse decision of selling too soon and watching a stock go up three, four, five times from where I sold it.Were you being sarcastic, or do VCs think it’s better to lose their money than to not make more money?

    1. Mark Essel

      I think this is truth. Selling a rocket ship too soon, can equate to a much bigger miss than riding a position to zero.

      1. cavepainting

        Interesting and very true. I do wonder if over a 10 year period, the return profile on any basket of public stocks would look similar to a VC return power distribution with the 80% of gains powered by 10-20% of stocks ?

  19. Dan Conway

    In my twenties was a buyer for a film studio. I’d read a spec screenplay or a manuscript out of New York and would have to make a decision sometimes in a couple of hours or overnight. I’d have one shot at it and then they were gone. Occasionally in the evening, I’d walk across the lot courtyard to my car with a bag of other scripts that needed to be read that night and run into the producer who I had outbid or had outbidded me earlier that day. A polite nod or “Good job on getting that manuscript today”. Competitive, sometimes cutthroat, but respect for the process.Career pivot and went into investment management in my late twenties. Took away tons from years in the film industry but “This isn’t screenplays or manuscripts” is one of the most useful. Today, I can be wrong but have the opportunity to go back, reevaulate and make another decison to get to right. Public companies will be there… tomorrow, next quarter, next year.

  20. Salt Shaker

    FW: “Selling is the hardest part”Tom Petty: “The waiting is the hardest part”Kind of agree w/ Mr. Petty. Patience indeed is a virtue.

    1. LE

      Agree with the waiting part being hard and you’d be surprised at how many people that I deal with and help (on the buy side) that I have to run interference on so they don’t panic and overpay because they don’t want to sit and wait things out. Along the same lines (on the sell side) I’ve had people that have made me offers that I was seriously considering that got impatient. They then made a higher offer (because I didn’t promptly get back to them which is often part of the game). In other words I hadn’t even rejected the first offer and then they bid things up.

  21. B Tenneson

    We had the same policy in our funds. We did not profess to be experts in publicly traded securities so left that decision up to our LP’s. Being consistent was worthwhile.

  22. jason wright

    i sell when i have something better to buy.

    1. creative group

      jason wright:despite all the fundamentals you use to guide you on an investment going opposite of your goal you would continue to hold it until you find another suitable equity? That is called ridiculous.

      1. jason wright

        it’s based on knowing that what i have to ‘sell’ has intrinsic value, and is not a plastic that will melt away.

  23. creative group

    NOTE TO SELF:The formulas, analysis and urgent need for the obvious on one equity is needed.The elephant in the room uses 140 characters. It doesn’t appear anyone is buying the rosy pitches to buy.

  24. george

    I was once advised, “you’ll never look like a fool for taking a profit.” If you want your business to survive, I think you you have to be disciplined to pocket profits when you can. I find it very difficult to control runway (founder/investor divergence); I rather manage factors that I share some relative control over – my own expectations.Having an investment/financial policy is extremely important, it helps me focus on the benefits (profits) vs. features (wealth). Choosing the right financial practice has helped me sort through the shades of truth when realizing net returns..

    1. Mark Essel

      I dig that quote, but sometimes feel like a fool for missed earnings. Being disciplined about trading would be a big breakthrough for me. I still trade on instinct and observation (poor to mediocre).For the most part its been ok compared to squatting on the S&P Index. Having safety triggers in place would be a nice improvement (stop orders). Selling partial positions after gains would be huge.

      1. george

        Thanks for the kind words…building a process comes with with a heavy dose of tears along the way…

  25. JLM

    .There is a difference between “selling” and either “de-risking” or “diversifying” assets. Selling is just the first step from a war zone to an oasis of comparative safety.When you are engaged in a risky business (VC is such a business), there is wisdom in moving assets into other classes (e.g. commercial real estate) wherein the risk profile is moving at a different wave length and amplitude.This could be justified solely based on the lower volatility of the alternative asset class.Risky businesses are, by their very nature, ones that can be unfavorably or favorably impacted by swings of sentiment — the definition of a highly volatile asset class — which may drive prices higher or lower but which will not be sustainable over some protracted period of time.People who make their money in highly risky businesses should store their wealth in less volatile asset classes. No huge revelation there, no?Deciding when to make the move is not nearly as important as actually making the move.I would also suggest that the management of the tax obligation of moving assets around is just as important as the asset class — meaning the ability to move assets through tax free exchanges of like kind assets isn’t chicken liver. This is tricky stuff but when coupled with skillful estate planning enables an investor to literally pay no taxes. Ever. Of course, you do have to die to make it work, so there is that. Sorry.Even within alternative asset classes, the nature of the asset itself may differ — hotels, apartments, CBD office, suburban office, office/showroom/tech, warehouses, storage (my personal favorite), retail, unanchored shopping centers, NNN leased single tenant properties, anchored shopping centers, mixed use developments, parking, and land.The inherent risk can be further mitigated by the nature of the management challenge — a hotel is an apartment building that rents and re-rents every night. Storage doesn’t have much active management and the ROI is 2-300 bp superior to other products because it isn’t sexy. They are easy to finance.One can take an active or passive approach — hire management or invest in “participating” mortgages which provide market upside with the security of a lien.People spend way too much time focusing on the exact second when they sell when they should be focused on putting risk derived money into substantially less risky asset classes as quickly as possible after a liquidating event.JLMwww.themusingsofthebigredca…

    1. Mark Essel

      Huh, wonder what it costs to begin investing in a storage business. Maybe something I can do with income savings ( < 100k) ?

      1. JLM

        .Buy a storage REIT.JLMwww.themusingsofthebigredca…

  26. David Pethick

    @fredwilson:disqus – any thoughts on how common (or not) this approach to exiting positions is for VCs?As one of the other commentators mentions, this is almost the only way to manage a position as a trader. On the other hand, I see lots of angel investors who look into the eyes of the management team to work out if they should sell or double down on their investment.I’m guessing that VC’s are somewhere in the middle – some quantitative measures, some gut?Dave P.

  27. cavepainting

    Great post !Fred, I would love to see a post on making the decision to sell out as a private company – especially in the middling scenarios where it is neither a home run nor a crash and burn.For example, a) when there is some promise of upside, but also a very real prospect of things getting worse, or b) big mission and great team with a lot of promise that has not come to fruition yet, but a real offer is on the table that does not pay a premium but provides an average return. In most of these situations, when the company has raised multiple rounds of financing, the founders and the different classes of investors are likely to have widely varying viewpoints in terms of what constitutes a good outcome.

  28. Mark Essel

    I keep thinking about this as I watch a personal Twitter loss. Is it time to get out or hold/go long. Stocks are a bit crazy, but also mix in some rationality (fundamentals bad for a long time = stock price falls).

  29. Blair Marshall

    I like your idea of having a non-emotional policy and selling at a point in time. I wonder how your returns would look, however, if you treated this like a second game of VC and instead always held on to every position for 5yrs, or 10yrs or whatever arbitrary time period your firm selects. My hypothesis would be, like VC, you would have a few big winners (FB, GOOG, etc) that make up for the losers and then some. I would imagine you would have a 3-5x return over that time period because the GOOG and FB of the world are returning you 10x over a 5 year period vs the 0x returns of the losers. On a weighted basis, you should have a meaningful return. Have you guys analyzed this method before?