5 ms·

"Maidenberg was frustrated by Palihapitiya's push to prioritize data in the VC investment process, which is traditionally based on a more high-touch, personal a

"Maidenberg was frustrated by Palihapitiya's push to prioritize data in the VC investment process, which is traditionally based on a more high-touch, personal approach." I think that is a telling line. There are certain decisions that can be improved with data, and many others where data can be your only criteria, but, in business, there are also a whole class of problems where data is basically a pipe dream. Don't get me wrong, I am not saying that wanting to make decisions with data is wrong, but, there is a lot of what goes into an investment that is "unknowable" in the sense that your "data" is very sparse or anecdotal, or, you simply have more unknown unknowns. It is this fact that causes many VCs to invest in good teams, that they feel will be able to handle the uncertainty, rather than investing in purely data backed opportunities.

subway 8y ago
Remember when the stock market was considered an art, dependent on your ability to make a deal? I'm sure a lot of those folks insisted the quants would never eat their lunch, because the data is a pipe dream.
- abakker 8y ago
  Personally, no, I don't remember that, but, I know it did exist. I'm also not saying that data won't ever exist. but, the stock market, as a market, produces the data needed for the quant funds. there is enough "price information" in the market to generate the data that a quant fund needs within the market itself. OTOH, startups are opaque, and most of the relevant data is kept out of the the hands of anyone that might aggregate it. instrumentation of the startup market does not benefit people the way that trading on existing data streams benefits the public ones. Again, not saying don't try to get more data, but, given the status quo, it is not a lot of data to go on right now.
- cirgue 8y ago
  Even the people saying the stock market was an art assumed you already had the data. Quants didn't master data, they mastered latency.
not_that_noob 8y ago
Index funds have beaten most active management over the long run. If you think about it, YC is a huge index fund in startups. The problem is not data-based prediction - just that it is impossible to pick winners a priori, no matter how much data you have. Ergo the clearly viable long run strategy is to invest in as many as you can. This is why YC is maximizing volume, with the latest being the Startup School. The other option of course is to invest large amounts into startups that are well on their way, but Sequoia and AH have a corner on that market. And now YC wants to play there as well, with their larger fund. So there's no oxygen left for a lot of the VC firms in the valley. There's no shortage of money looking for VC funds, as returns are low in general asset classes. So you have this imbalanced situation, where dinky startups get incredible valuations, and there are too few genuine investment opportunities to go around. SC are one of the marquee victims, but other walking dead abound, though they dress snappy so it's hard to tell they're really dead.
- pmart123 8y ago
  For startup investing the pure math explanation to invest in 1,000 companies makes more sense than in public markets. If you have 1 out of 1,000 startups return 1000%, 99 companies break even, and 900 companies go bust, you still make money due to the asymmetric one company generates. In public markets, two deviations of risk is satisfied with 20ish companies where three deviations being around 80 companies assuming a log-normal distribution. Adding more companies does not greatly lower the risk when the outcome isn't so binomial.
  - mmt 8y ago
    I assume you meant 1000x not 1000% (since the latter would only be 10x. Is there any evidence that even as many as 1/1000 startups provide a 1000x return to VCs, considering that the return is only a fraction of a total exit? Even then, assuming all investments were equal, that's a very meager return of 1099 on 1000. For a 10-year fund, that wouldn't even beat inflation, let alone compete with something like Treasuries. An article that delves into more details (although there's some glossing-over, still): https://news.ycombinator.com/item?id=17874278 https://news.ycombinator.com/item?id=17874278 https://techcrunch.com/2017/06/01/the-meeting-that-showed-me-the-truth-about-vcs/ https://techcrunch.com/2017/06/01/the-meeting-that-showed-me...
pmart123 8y ago
In Palihapitiya's defense, he presented a logical argument of how this could work when I heard him speak in person. Essentially: 1. More SaaS products like using Stripe for payments, AppAnnie, etc. can audit and monitor revenue. 2. Small, seed stage investments up to 250k and placing a lot of bets. 3. Companies could apply online allowing for startups in non-traditional and underserved areas like Mexico City to receive funding. Ditto as well for removing potential human biases like founder gender, race, etc. I don't believe he ever proposed using a completely data driven decision for a $200M Series C round. Personally, it seems like an unfortunate situation as I like when someone undertakes a new approach to an old problem, but it sounds like he stopped showing up to do the work needed for this to happen.
matchagaucho 8y ago
I don't believe such a data-driven approach could avoid the bias of age, socio-economic background, gender... the list goes on. Start-up "Disruptors" have a lot more immeasurable hustle.

subway 8y ago

Remember when the stock market was considered an art, dependent on your ability to make a deal? I'm sure a lot of those folks insisted the quants would never eat their lunch, because the data is a pipe dream.

abakker 8y ago
Personally, no, I don't remember that, but, I know it did exist. I'm also not saying that data won't ever exist. but, the stock market, as a market, produces the data needed for the quant funds. there is enough "price information" in the market to generate the data that a quant fund needs within the market itself. OTOH, startups are opaque, and most of the relevant data is kept out of the the hands of anyone that might aggregate it. instrumentation of the startup market does not benefit people the way that trading on existing data streams benefits the public ones. Again, not saying don't try to get more data, but, given the status quo, it is not a lot of data to go on right now.
cirgue 8y ago
Even the people saying the stock market was an art assumed you already had the data. Quants didn't master data, they mastered latency.

not_that_noob 8y ago

Index funds have beaten most active management over the long run. If you think about it, YC is a huge index fund in startups. The problem is not data-based prediction - just that it is impossible to pick winners a priori, no matter how much data you have. Ergo the clearly viable long run strategy is to invest in as many as you can. This is why YC is maximizing volume, with the latest being the Startup School. The other option of course is to invest large amounts into startups that are well on their way, but Sequoia and AH have a corner on that market. And now YC wants to play there as well, with their larger fund. So there's no oxygen left for a lot of the VC firms in the valley. There's no shortage of money looking for VC funds, as returns are low in general asset classes. So you have this imbalanced situation, where dinky startups get incredible valuations, and there are too few genuine investment opportunities to go around. SC are one of the marquee victims, but other walking dead abound, though they dress snappy so it's hard to tell they're really dead.

pmart123 8y ago
For startup investing the pure math explanation to invest in 1,000 companies makes more sense than in public markets. If you have 1 out of 1,000 startups return 1000%, 99 companies break even, and 900 companies go bust, you still make money due to the asymmetric one company generates. In public markets, two deviations of risk is satisfied with 20ish companies where three deviations being around 80 companies assuming a log-normal distribution. Adding more companies does not greatly lower the risk when the outcome isn't so binomial.
- mmt 8y ago
  I assume you meant 1000x not 1000% (since the latter would only be 10x. Is there any evidence that even as many as 1/1000 startups provide a 1000x return to VCs, considering that the return is only a fraction of a total exit? Even then, assuming all investments were equal, that's a very meager return of 1099 on 1000. For a 10-year fund, that wouldn't even beat inflation, let alone compete with something like Treasuries. An article that delves into more details (although there's some glossing-over, still): https://news.ycombinator.com/item?id=17874278 https://news.ycombinator.com/item?id=17874278 https://techcrunch.com/2017/06/01/the-meeting-that-showed-me-the-truth-about-vcs/ https://techcrunch.com/2017/06/01/the-meeting-that-showed-me...

pmart123 8y ago

In Palihapitiya's defense, he presented a logical argument of how this could work when I heard him speak in person. Essentially: 1. More SaaS products like using Stripe for payments, AppAnnie, etc. can audit and monitor revenue. 2. Small, seed stage investments up to 250k and placing a lot of bets. 3. Companies could apply online allowing for startups in non-traditional and underserved areas like Mexico City to receive funding. Ditto as well for removing potential human biases like founder gender, race, etc. I don't believe he ever proposed using a completely data driven decision for a $200M Series C round. Personally, it seems like an unfortunate situation as I like when someone undertakes a new approach to an old problem, but it sounds like he stopped showing up to do the work needed for this to happen.

matchagaucho 8y ago

I don't believe such a data-driven approach could avoid the bias of age, socio-economic background, gender... the list goes on. Start-up "Disruptors" have a lot more immeasurable hustle.