Loss Ratios In Early Stage VC
When I was early in my career, I casually mentioned to an older VC that I had yet to lose money on an investment. He replied "that's not good, you aren't taking enough risk." I have gone on to lose a lot of money over the years. And made a fair bit too.
So one of the things I like to look at when I look at our funds and other VC funds that I am an investor in are loss ratios. You can calculate loss ratios by "names" meaning how many investments ended up being worth zero. Or you can calculate loss ratios by "dollars" meaning how much of the capital invested in the fund went into total losses. Ideally your names loss ratio will be a lot higher than your dollars loss ratio.
Our first fund, USV 2004, has an "names" loss ratio of about 40%. That means 40% of the investments we made are going to be end up being worthless or near worthless. That fund will be the best venture fund I have ever worked on. So loss ratios are not really indicative of performance of a fund. That comes from the winners and how big they are.
I just looked at the financial reports for a seed fund I have an investment in. It has a 42% "names" loss ratio three and a half years in. That sounds about right to me. I feel good about that fund. If it had a lower "names" loss ratio, I might not feel as good about it.
I am not suggesting that a high loss ratio is indicative of good performance. It is not. But it is indicative of risk taking, and importantly, taking your lumps and moving on. The worst thing you can do in early stage VC is stick with your bad investments for too long and for too much money. If the "dollars" loss ratio is higher than the "names" loss ratio, that can well be an indicator sticking with your losers too long and that can be an indicator of poor performance.
The point of all of this is losing money comes with the territory in early stage VC. It is not something to be ashamed of. But it is something you need to be doing as quickly as you can.
Comments (Archived):
Man, does this post hit home. I used to say if I batted over .300 on trading, I’d be in the Traders Hall of Fame. I think VC is the same. I journal all of my investments. Some die because I thought they were bang up ideas that weren’t; or the startup was unable to raise enough capital to get going; or poor execution; or events beyond your control (2008 for example).The key metric for me is the way I make decisions. The hardest ones are the next rounds. Do I re-invest or not? Why and why not? One of the smartest decisions I ever made was not re-investing when the crowd did. Another was reinvesting when the crowd didn’t. Preservation of capital is sometimes a better decision than deploying it.It’s never fun to lose money. One of the reasons I can’t go to Vegas. Enough risk in what I do.
A lot is when you get your hits too. 2 guys on a shot to the wall is great even if you are 0 for 4 at that point. Timing and I always a bit a luck
how do you feel when you reread your journal – must be interesting…
So, the “names” loss ratio is a measure of the degree to which the VC is failing fast, i.e., learning from early mistakes and course correcting. Seems to me then that this is a key metric for judging performance of 1st time managers and VC teams.
You only have to be right 60% of the time? Maybe even less. Interesting.
I think it is <20% & could be close to 10.He needs to be ‘not wrong’ (break even) 40-50% of the time…
Yup the 3x+ returns drive the whole portfolio for an early stage fund – see the data I posted above. A VC needs to only be right 10-20% of the time – if right is defined as 3x+ returns on a company – to beat the averages of the past decade of venture investing.
VC only really needs to be right once a fund.It is the cash management (cutting losses, maximizing winners) that defines the other 50% of this hockey saying:’goaltending / winners is 50% of the game, unless you don’t have it – then its 100% of the game.’
Years ago, I asked Joe Sugarman, the crazy direct marketer (remember blueblocker sunglasses) about his money back guarantee.I wanted to know how often people asked for their money back.He said, “If it’s less than 10%, I didn’t write the ad boldly enough.”
joe is awesome. he pioneered taking credit cards for remote transactions (ie phone and mail). back then it was forbidden to charge someone’s card without a signature. he just went ahead and did it. finally the card companies changed the rules.
what are bluebocker sunglasses?
There is a direct correlation, in general, between how overpriced something is and the money back guarantee offered.I don’t mean “overpriced” pejoratively either.I mean you are able to raise the price in order to offer a worry free experience on the hope that an increase in sales offsets the nudniks that want a refund. So it’s like insurance. In the end you make out.Also true with auto renewals and extra charges. Better to charge someone and refund than to ask before charging. (Stripping out of course customer satisfaction issues). [1] I know everyone is going to take issue with this statement and yes it does vary depending on the business in particular and what you are doing. I’m make sure to put all the details in the book that I’m not going to write.[1] For the life of me I’ve never understood why restaurants tell you that “it’s going to be $1 more for that” when they could just charge “$1” and then cheerfully refund anyone who notices and complains. I’m guessing that there must be a dynamic that I haven’t accounted for or just that they are scarred by whinny customers and shouldn’t be. I remember way back when we instituted a delivery charge (in a business I had) which was something like $5. Never told the customer just started one day. Practically nobody complained and the ones that did just mentioned it nicely. Instant revenue boost. Just like that. And for really good customers we said “oh sorry not for you” and it was better than if we never tried in the first place. Everyone wants to feel special.
If you look at any infomercial stuff, and you want to know COGS. See what S&H is. That is the number.
These days, I’m one of the few people I know who’ve read Joe Sugarman.
Interesting….For the investor, its averages.For the startup entrepreneur, it’s life and when you end up in the loss bucket, lots of time and pain.Life from different angles has different quality scales.
Investor —> DiversifiedEntrepreneur —> All eggs in one basket
Investing –> diversificationCompanies –> eggs in on basket(Pretty sure a noble prize was given for this?)Interestingly – Portfolio theory does not seem to apply to VC.
Portfolio theory does work in VC. That’s why a fund shouldn’t have one investment only. One should add to the portfolio more companies if one is able to have the same information about them and and not reduce the added value with the increase of companies in the portfolio.What I think you were saying is that in VC it’s not about having as many companies as possible to diversify the risk, but that’s for the reasons stated above. When I mean “information about the companies” I mean having the right knowledge about the project, the team, and the field.I think VCs end up being a pre-filter to the portfolio basket (and thus do select the investments – i.e., not random investments), but after that, the higher the diversification the better. I would even add that, from a theoretical point of view, one would best invest the same value in first rounds in every company (in order to maximize diversification), and they let the bad ones die, and double down on the ones with business traction.
Diversification within a sector != diversification (portfolio theory would argue that usv have no investment themes.) See @sigmalgebra’s comment for the proof.
My argument with independence, the samedistributions for all the return random variablesX(i), finite expectations E[X(i)] (easy to assume inpractice), and the weak law of large numbers will,right, fail for USV’s portfolio based on “largenetworks of engaged users”. Mostly what fails isthe independence assumption.However the Markowitz argument without myassumptions of independence, etc. but with variancesand covariances known (and likely want to assume tobe finite) still holds for USV’s portfolio. Alas,as we would guess intuitively, the returns in USV’sportfolio (all “large networks of engaged users”)might be highly correlated, and the result from theMarkowitz work promises to be, for any selected,possible expected return, the lowest possible riskhigher than with my rather strong assumptions.Yes, the Markowitz work would say that an investorshould look for relatively uncorrelated investmentsor, perhaps, even negatively correlated investments.
Depends on how you look at it. Spray and pray is not portfolio theory. Investing as a VC automatically assumes you have more information than the market. So, you have a theoretical unfair advantage by using asymmetric information. Investing and utilizing your network should give you a chance at arbitraging it for outsize gains. Make enough investments, and odds start to go your way. EMH assumes all information that is public is included in the stock price. That’s why inside trading is illegal.
Good point on the asymmetric information.
Noble indeed. Maybe even Nobel.
First day of the week, first spelling error. It may be a long week.
Yes, and both are counting their chickensbefore they are hatched?For portfolio theory, the basics are simple andstill apply fully: Suppose for some positiveinteger n we make n investments. Supposefor i = 1, 2, …, n, investment i yields returnX(i) which we regard as a ‘random variable'(intuitively clear enough but with a reallynice, solid definition from Kolmogorov –look in books by M. Loeve, J. Neveu,L. Breiman, K. Chung, etc.) that takesvalues in the real numbers (in K-12,the ‘number line’).Then for a simple case, suppose theX(i) are independent and all have thesame distribution, an expectation (i.e.,’average’), and, thus, the same expectationE[X(i)]. Then from just the weak law oflarge numbers (easy to prove with justelementary techniques), the expectedvalue of the ‘portfolio’ is just nE[X(i)]but with variance that, as n increases, goes down like 1/n and, thus, standard deviation that goes down like 1 over square root of n.So, the portfolio with the ‘diversified’investments does nothing for theexpected return but reduces the variance (i.g., ‘risk’) of the return.That is, diversification lowers risk.But if drop the assumption of independentwith the same distribution but know thevariance of each X(i) and that it is finiteand know the the covarianceof each X(i), X(j) pair, then can use asimple quadratic, convexity argument,as H. Markowitz did for his Nobel prize,to show how to allocate the investmentcapital across the n investments so thatfor any possible expected return froma portfolio of those n investments canget the least possible variance of thatreturn, that is, the least portfolio’risk’.For ‘modern portfolio theory’, i.e., fromW. Sharpe as in his Nobel prize (thereare books by Sharpe, Luenberger, etc.),to consider prices that result from a’market’ where each investor knows thevariance and covariance numbers.Partly, of course, this is some quitesimple mathematics that gets somereally clean results from assuming more data than really can hope to havein practice. Still, this is some reallygood ‘applied’ math, e.g., has resultedin more than two Nobel prizes!Once I pulled a trick something likethat: In grad school, I did someresearch and got some nice results,in part somewhat amazing. WhenI went to publish I discovered that I’d also solved a problem stated butnot solved in a famous paper inmathematical economics by Arrow,Hurwicz, and Uzawa. Poor, Uzawa:last I heard he hadn’t gotten hisNobel prize yet! Neither have I! Butmy work really was ‘applied’ mathif only because it was the last Ineeded for a Master’s!Somewhere back in English lit classthere was a line something like”Of the writing of books there is noend.”. Or we all know what BSis. MS is more of the same, andPh.D. is piled higher and deeper?
it should….
Unlike portfolio theory, there is no way to lower risks. My hunch is that the number of investments by usv in not at all motivated by diversification.
Likely could assume that the USV investments are ‘conditionally independent’ given that they are allfor “large networks of engaged users”.Then conditional independence shouldbe enough to get the traditional effectsof diversification, of course, conditionedon “large networks of engaged users”.That is, if all investments in such networks go to 0, then so will the USVones. Generally diversification of theUSV portfolio should move the USV returns closer to the average returnsof all investments in such networks.Note: I have yet to write out the math,so the above is an intuitive guess.
USVs performance is picking assets with superior expected returns not managing risks. Correlations if anything make their portfolio less efficient not more
Generally the Markowitz-Sharpe workis what to do when have more data thanis reasonable in practice, something likeif pigs could fly then should invest in castiron umbrellas.Strong negative correlations, if known,could help: Suppose with investments A andB, each has very high expected return butcorrelation of -1. Then bet on both, writeoff the loser, and take the winner to the bank.Right, risk reduction in VC is an awkward fit.As a startup entrepreneur, with nearly all myeggs in just one basket, my approach to riskreduction is to try to do good work!
Fred, and most VCs, do not, even in the rare cases that it exists, invest in negatively correlated companies. Why? The companies would likely be competitors. The mean-variance bullet has no meaningful relevance to the VC world.
Your second statement goes too far. All other things being equal, the standarddeviation of return per million invested willbe lower for a larger fund, and that will help keep the LPs happy.
Life –> All eggs in one basketErgo Startup –> Roller-coaster !
That Ergo sounds like a bumpy ride. Will I toss my cookies?
Not cookies – Omelettes – you can’t make them without breaking the eggs! 🙂
The entrepreneur survives beyond a failure and still owns 100% of their own talents and experience.
Certainly.
I’ve come to believe that there are many great ideas that will grow into great companies that are not VC material. That is not a slight, that doesn’t mean they are somehow not great ideas.I’ve run the numbers all sorts of ways and if you are investing other people’s money and making money from that, you have to set it up the way its been setup.If its your own for you, you can make different investments, but if its institutional you have to believe: “The worst thing you can do in early stage VC is stick with your bad investments for too long and for too much money”The pain comes in when you convinced somebody that has to have this attitude when your company does not fit that mold.That is when time and pain happens. Sometimes you just end up there even though you didn’t mean to, and that hurts.
Sadly I see many of them taking VC money. The Everpix story that was kicking around is absolutely one of them. This was a case of taking and spending too much money instead of living on decent revenues from customers.http://www.theverge.com/201…
Good read as always. What struck me was the size of the AWS bill. When there is a gold rush, its always really good to be selling pick-axes and pans. Searching for gold? Not as good.
Spending too much money supporting the free customers in the freemium model. That’s a pretty straightforward tweak.
yup, but I still don’t understand how they let it be that bad. Or why many vcs turned down revenue over users…
When you have revenue you have metrics.
This is also a good read on the same topic: http://subimage.com/blog/20…
the combo of a great idea plus amazing tech is not always a winner. somewhere in there strategy plays a role, this is the bit often overlooked -and hardest. today building the tech is pretty much a given, it’s not the risk is was. today, strategy is the most rigorous challenge, was less so in pre AWS / RoR days.
Disagree. When you are spending $35k a month on AWS….
business strategy #fail that incorporates a math #fail. AWS was a known variable, and an easy one at that. it’s like a parent ‘forgetting’ their kid’s tuition is going to be due on a certain date (i.e, buying a tesla) and not being on a sober critical path for building savings. a good critical path generally comprises at least one business strategy -and optionality. not being judgmental at all, looking at the AWS figure drives only home the point.
I agree.I have one current project that I may shut down and restart. My deal, I just do it.I have another that is gaining real traction, not a tech company but needs capitalization and considering simply a small business loan.Others though, are in the VC threshold and I”ll go that route.
“and considering simply a small business loan.”Small business loan just solves the cash flow problem. Meaning you are on the hook if you can’t pay it back. It might as well be money you take from your bank account. [1]Taking investment means you aren’t risking your own money. In exchange you have to answer to somebody. But the truth is we all eat shit just at different tables.I like the fact that you are diversifying and working on what appear to be several fronts at the same time.[1] I’m not up on any SBA type stuff I’m talking about traditional asset based lending or where you sign personally.
This is a litmus test to tell a novice, even a bad, VC: when they tout their loss ratio of zero to LP’s, require downside protection on their early stage deals, you can tell that they have a PE (not VC) mindset.The only exception if VC’s help entrepreneurs get on base, assuming that entrepreneurs are interested in respectable and not just moonshot outcomes.Loss ratio minimization strategy is bad if as a VC you do it for your LP’s, but may be good if you do it from a perspective of helping your entrepreneurs.
Is it a good thing if the names and dollars loss ratios are about in the same % vicinity?
Number of users weighted score =Sum i : 1:N ( investment (i) x number of users (i) / sum of investments)Needs work 🙂
Gonna HAVE to introduce you to D. Knuth’sTeX for how to type math!sum_{i = 1}^N x_i n_i / left( sum_{i = 1}^N x_i right)The left, right pair makes the parens big enoughto cover the content inside the pair. The underlinecharacter _ starts a subscript. Curly braces arefor logical grouping. E.g., the i = 1 is treated asa ‘subscript’ on the summation sign sigma denoted bythe sum. Yes, the backslash is an ‘escape’character, but tweaking that is possible.Of course there is Knuth, ‘The TeXBook’. With a lotmore macros and a slight tweak to the basic TeX getLaTeX where the documentation is larger and lesswell written.TeX was supposed to be a way to use computers andlaser printers, later PDF files, to ‘typeset’traditional mathematics that before had been donelargely by hand. TeX was not supposed to be agrand, new way of handling all text for the future,e.g., the Web, advertising copy, high end graphics,etc. For what TeX was intended for, it’s excellent,that and more.There are free software distributions on theInternet, say, at CTAN — comprehensive TeX archivenetwork or some such. TeX output goes to a ‘deviceindependent file’ (three letter extension DVI), andthere are video ‘preview’ programs that will displaythat nicely. Also there are programs to convert DVIfiles to PDF files, at least for the fonts that arestandard with TeX.The Web has made some efforts to support mathnotation, but what Knuth did was so well andcarefully done that it will be tough to replace it,at least in the quality of the output (e.g., finedetails of spacing).In principle, TeX can use any fonts you want andeven has some quite nice software from Knuth,Metafont, for creating new fonts. In practicepeople mostly stick with the standard fonts fromKnuth, a few other sources, and the AMS (AmericanMathematical Society) collection of mathematicalsymbols.The ‘program logic’ documentation from Knuth for TeXand Metafont are ground breaking and exemplary.TeX and LaTeX are now the essentially unchallengedinternational standards for typesetting nearly allserious work that makes heavy use of mathematics.For the actual typing, for ‘text style’ output,intended within ordinary text, use ‘$’ as a prefixand suffix as in$sum_{i = 1}^N x_i n_i / left( sum_{i = 1}^N x_i right)$For ‘display style’ output, intended to be on linesof its own, use ‘$$’ as a prefix and suffix as in$$sum_{i = 1}^N x_i n_i / left( sum_{i = 1}^N x_i right)$$Now you have had TeX 101!Correction:After seeing Cam MacRae’s post and trying with TeX,changed )right to right) as in:sum_{i = 1}^N x_i n_i / left( sum_{i = 1}^N x_i right)
$ $[ sum_{i=1}^{N}frac{f(i)g(i)}{sum_{i=1}^{N}f(i)} ]Doesn’t work. Would have been delighted (and surprised) if it did.
Ah… I see you updated your comment. Obviously no mathjax on avc.com 🙂
Thanks for your remark. See my revised post and my replythat has some graphics output.
Nice. I prefer yours stylistically.
Here is the display output for TeX lines$$sum_{i = 1}^N x_i n_i / left( sum_{i = 1}^N x_i right)$$$$sum_{i = 1}^N {{ x_i n_i} over { left( sum_{i = 1}^N x_i right) } }$$bye
You are absolutely right, Fred.For early stage VCs: loss ratio is like martinis – a moderate amount is OK but too much is bad. Too little is no fun.According to Cambridge Associates, in the post dotcom era (2002-13), 56% of the 4,169 realized VC exits in their dataset (US companies, US VC funds) have been < 1x return. 7% have been > 5x in the same era.The full data 2002-2013 – realized exits:Full loss: 17%Partial loss (>0x <1x): 39%1x: 2%1x-2x: 17%2x-3x: 9%3x-4x: 6%4x-5x: 3%5x+: 7%Fascinating to see how this has changed from the 90sThe full data 1990-1999 – realized exits:Full loss: 23%Partial loss (>0x <1x): 35%1x: 1%1x-2x: 11%2x-3x: 7%3x-4x: 5%4x-5x: 3%5x+: 15%In the 90s: more total losses (no acqui-hires) and almost 2x more 5-baggers.Ahh the good old days.Edit: by the way – this is done on a “names” basis and not dollar-weighted and across all stages of VC investment. I may update later for early stage only.
Thanks for sharing this data. I had not seen it. Very helpful
I saw some data from a StepStone SPI database courtesy of Prequin that said fund size is also a predictor of returns. Funds between 50-250 are 5x more likely to return over 2x net invested capital to LPs and 10x more likely to return over 5x invested capital.
i wrote a post about that a long time agoi can’t find it right nowwe have raised four funds between $125mm and $200mm and are raising two more right now between $150mm and $175mmi think the sweet spot is $125mm to $150mmabove $200mm and you have the problem that you need a $1bn to $2bn valuation company in your portfolio to hit your expected returnsand that is hard to deliver every three to four years consistently
this makes sense. every time i hear about a new super huge fund… i think of this exact issue.
I mean… we are absolutely going to be a $B company… just saying…for like…your average company.
In addition, combined with what you are saying this probably relates also to a people issue. How many partners in the VC fund. Meetings, managing investments, boards, entrepreneurs. Daily shit that takes time to do. How many things can 1 person or 5 get done and still pay attention and do the right thing? (You have already highlighted how you expanded years ago and decided to keep small.)Otoh A girl that I dated her brother was in hedge funds. He had a small 100 million fund. He said if he raised more he would just make larger bets on the same stuff. So it scaled pretty well if I understood his point.
Is there a web url for that data or is the source private? thanks.yup re: martinis. 1 is not enough, but 3 is too many already 🙂
Unfortunately from a private database they sell
From a southern grand dame:” I like to have a martini, two at the very most. Three I’m under the table, four I’m under my host.”
ha…
care to name names for that brilliant gem of a quote?
Dorothy Parker.
I never knew my wasband’s granny was quotin’ Ms Parker, as pointed out by @cammacrae:disqusFranny was of that era and all tripped off her tongue like that until she was 101.5. Ask her ” Franny, can I get you another drink?” and hear back “I thought you’d nevah ask…”
Commander’s Palace, New Orleans. Lunch, 25 cent martinis, limit three. : )
I was there…way back when Emeril Lagasse was the chef. But I don’t remember to 25 cents martinis. Is that still going on? You might drink them fast because of the food’s spicyness 🙂
Must have been well after I dined there. Great restaurant, in fact one of my all-time memorable meals.
Thanks. Link?
“a seed fund I have an investment in. It has a 42% “names” loss ratio three and a half years in”:I’m just guessing that we could multiply that times 500 startups?:)
How many VCs in the current environment would have stuck around with Twitter from 06 until 09?
With the user growth? Lots.Twitter is really good a defining metrics that look terrific. Now that they are public there are a different set of important metrics.
Twitter’s user growth was pretty much flat from 06-09. that’s my point. Those that stuck with their guns did well. http://www.google.com/trend…
It depends on the scale of the graph.http://www.google.com/trend…
fair point!
Cue the idiotpreneurs walking into VC offices with this post in hand.”Trust me, I am going to be one of the losses that MAKES your fund.”
I would love to do that just once, maybe on April 1st. Have a beautiful deck that just shows me putting a smoking hole into the ground. We’re going to have the best offices with the best games, we’re going to throw pre-launch parties the likes of which you’ve never seen!You know you need 40% of your investments to be totally worthless. We are going to do that with a capital W and an explanation point!
This really needs to be a hidden camera show. “Undercover VC”
Or Punk’d.
The idea with this show would be to see how long the person pitching could keep going until they were finally kicked out of the office by the VC who knew they were being punkd.On Shark Tank one of the strategies is to stretch things out long enough (regardless of success or failure in fund raising) so that there is enough footage to get broadcast. Like with prize fighting nobody wants a knockdown in the first round.
sadly you might get a check.
Fred, recognizing that unicorns will only be built by taking outsized risk, why can’t there also be a school of thought supporting doubles & singles at lower risk. If we’re aiming for a 3x fund return, can’t we be focused on EV (Expected Value) rather than unicorns?For example, on Instagram I might calculate my EV as:$1BN * 1%$500Mn * 2%$250Mn * 2%$100Mn * 2%$15Mn (aquihire) * 30%$0 * 63%EV = $33MWhereas another deal might have EV of:$100M * 15%$75M * 15%$50M * 30%$25M * 25%$10M * 10%$0 * 5%EV = $49.75MSo b/c the distribution trends far less towards $0 even with a lower cap my risk adjusted EV is higher. Just curious what your thinking would be on this and whether you would back the 2nd company if you got in at $15M post…Ezra
The problem is you are saying there is an 85% chance of a $25mm+ exit.That’s not even realistic if you currently have a letter of intent in hand from a real company for a $25mm exit.
I mean it’s a theoretical question on EV vs chasing unicorns…
The problem is that the numbers just don’t work. The reason I like this post is that it is near and dear to my heart.After an exit, years ago I really thought: Why can’t I have a VC fund that concentrates on lower risk/lower return ideas. No 10xers. Those are the companies I like, I don’t understand unicorns.Well if you work the numbers, you put in you will do small rounds and then follow on etc, etc, the bottom line is to make your numbers on a portfolio that includes paying yourself and giving a return that investors will be happy with you need investments that return the whole fund. There is no way around it.
I get it – I just wonder if there’s a way to optimize for investments with more normal distributions. But Fred’s point is good that it’s doable at growth stage but not early.
math is underrated
very underrated
The real issue is why VC returns net of fees and risk adjusted keep lagging the S&P and Russell 2000, as Kaufman report and, to his credit, Fred Wilson, have written about.You have Harvard Management pretty openly complaining about how they can’t get their money out of VCs despite their own models showing that VCs are not good investments these days.Where’s the alpha?
exactly
i have not seen anyone make this model work at the early stagethis is exactly the model growth stage VCs usei think the problem with trying to hit singles and doubles in VC is you still strike out just as much
Great, thanks for the insight across the early stage funds you’ve seen that failure % still as high. Will internalize.
It’s more like you’re swinging and hoping to connect with a pitch that was already moving towards the outfield in the first place 😉
why is that early stage you strike out so much
Aha!
Where’s the data they strike out just as much? Thanks.
Gee, weren’t we all told that 99 44/100% of all business startupsfail?
All startups Fail – Some make money for some investors in the interim.All people die – Some make interesting decisions in the interim
Risk taking VCs have to maximize the standard deviation for the performance of their investments.
i went shopping today. just got back. bought a pair of Bont A3 road cycling shoes (a startling footwear experience) , and then i came across this anti conundrum;google nexus 7 tablet GB32 with cellular – GBP 299 apple ipad mini retina GB32 with cellular – GBP 499this is a ratio to ponder.
I did some analytics in the General Brokerage insurance industry for a while.It is possible in insurance to make profit on loss-making lines of insurance, if investments on held premium exceeds commissions and operating expense.The implication is that insurance is the opposite of venture – which makes sense.However Loss ration always refers to money not “names”, which would be the claims frequency.The biggest reserve of insurance is IBNR (claims incurred but not notified) a sort of cohort analysis allows reserves for as yet un-notified claims to be held, and naturally tax is not due on such reserves as they cover liability.I wonder if a VC could claim a non-profit reserve on future name losses that have been incurred but nobody noticed yet ! (For some VC this may exceed their fund !)
Based on that wise VC’s counsel, I can safely say I’m taking enough risk in life.
When I was early in my career, I casually mentioned to an older VC that I had yet to lose money on an investment. He replied “that’s not good, you aren’t taking enough risk.”I think the fact that VC’s take risk and that that can be quantified and related to success or how much money they end up making is not the same as saying that if you don’t have failures you haven’t done your job (which is what it seems the old timer was saying). (Inverting the thinking) because you haven’t bet big enough.What I think it does is it gives someone a convenient excuse for screwing up. As if to say “hey no problem with that failure is an option”. To much of this already. Like “there will be infections in hospitals just the way it is”.Because what it does is not factor in the reasons for the failure (or lack of it) but just paint a broad stroke by looking at numbers (or binary outcome) which to me is just a lazy way of doing things. Although clearly things are done this way.I remember many years ago a time when I had very little a/r losses of any significance. I concluded that I must not have been extending credit when I should have (which more or less dovetails with what the older VC was saying). And that if I did I might have ended up with more profit. But what if I had simply expected a certain amount of bad a/r without thinking any more than that? So if I had bad debt of 7% and said “oh that’s ok it’s expected” instead of “why is it even 7%?”.I guess what I am reacting to is just the simple pat answer that the VC tried to guide with not the fact that you couldn’t derive some significance from looking at things that way. To much (by the way stated) to the “spray and pray” angle and not down to specifics which would actually make a difference.
Great comment.
If you ever find yourself too low on that ratio…come to me…I’ve discovered *lots* of ways to lose money over the years…I’ve been blogging about many of them in my didn’t work wednesday series each week. 😉
Longtime listener, first time commenter. We did 23 investments in our first fund. 10 will go to zero or were picked up for pennies on the dollar. 43%.I love how coincidentally close that is to your fund and the fund in which you invested.
There’s nothing magical about 43%. Without seeing a lot more data, this is just a made up number.It really feels like yet more Valley group think.Might be time to short the NASDAQ, the closest way I know of to synthetically short startups.
As I’ve told younger colleagues – “Everyone makes mistakes – Rule #1 is to learn from your mistakes”
Higher loss ratios are only good if they actually lead to superior risk adjusted returns net of fees. This seems like another b.s metrics that barely correlate with actual performance.This kind of startup math reminds me of VC group think during the 90’s about eyeballs and pushing startups to spend more money.An LP doing its job should arguably consider terminating the services of any VC that can’t deliver long term risk adjusted returns net of fees > the average VC .An LP’s only job is to find VCs that can deliver alpha.Anyone can match or underperform the market.Risk adjusted returns net of fees. The rest is noise.
Loss ratio should be proportional to how long the VC has been in biz. Sure biz, technology etc changes over time, but experience along with asymmetric info should offset that. Its like, I as an investor can make mistakes on 40% of my overall investments, but same cannot be said of Warren Buffet!ToadsterPS: Just to clarify, 40% of the portfolio companies are doing poorly, the the remaining 60% more than offsets the loss in $$ and hence the investment is ahead
our first fund raised in 2005 (which is a top performing fund)had 50% strike outs with big hits on Diapers.com, Yodle, Payquik. I’ve tried to stop swinging at pitches “out of the strike zone” but that might cut down on overall returns.
Thanks for giving me the useful information. I think I need it. Thank you
Wonder how investing in conflicting companies (a la SV Angel) changes the percentages? Not sure about the relative weighting of segment risk versus company risk.
Hm, that’s almost 50:50, i.e. pure guess/”blondes or monkeys with darts” scenario… Why bother with DD in that case?.. I wonder what the upside average is, though…
The higher the risk, the higher chances of earning big. But always be cautious, loss always come halfway.