How to size bets
Some personal principles 👇
Some personal principles 👇
1/ Picking the right investments is half the game; sizing the bets is more important.
2/ A wrong bet sized right can be a valuable lesson; sized wrong it can spell absolute ruin.
A right bet sized wrong means you worked for nothing; sized right means victory.
A right bet sized wrong means you worked for nothing; sized right means victory.
3/ Your framework for sizing should optimize for upside while limiting tolerable downside.
Tolerable because everyone's appetite for risk varies.
A hedge fund may have investors redeem if they end the month down 20%; a prop guy might eat 20% days for breakfast.
Tolerable because everyone's appetite for risk varies.
A hedge fund may have investors redeem if they end the month down 20%; a prop guy might eat 20% days for breakfast.
4/ In bet sizing, you can generally manage for risk (top down) or manage for upside (bottom up).
Top down: Limit every bet to x% of book
Bottom up: Size every bet based on conviction
Top down: Limit every bet to x% of book
Bottom up: Size every bet based on conviction
5/ Good outcomes != good process
Just because your friend printed millions by aping their net worth into 1 shitcoin doesn't mean you should trust them with your money.
In the investing game, performance OVER TIME is the only truth.
Just because your friend printed millions by aping their net worth into 1 shitcoin doesn't mean you should trust them with your money.
In the investing game, performance OVER TIME is the only truth.
6/ Example of top-down sizing: only making standard check sizes, regardless of conviction or bankroll.
Common in: angel investing, spray and pray funds, accelerators.
Common in: angel investing, spray and pray funds, accelerators.
7/ Example of upside-based sizing: only making an investment if you think it can return the fund.
Common in: funds with discipline around fund size and valuation, ok with missing small wins in order to prioritize giga home runs (h/t @richardchen39)
Common in: funds with discipline around fund size and valuation, ok with missing small wins in order to prioritize giga home runs (h/t @richardchen39)
8/ Example of EV-based sizing: using Kelly criterion to determine optimal bet size.
Bet size = p - (1-p)/b
Where p = probability of win (subjective, or based on historical perf), b = odds (e.g. in a 2-1 odds, b = 2)
Bet size = p - (1-p)/b
Where p = probability of win (subjective, or based on historical perf), b = odds (e.g. in a 2-1 odds, b = 2)
9/ Kelly is common in poker, and some VCs where risk of an investment going to zero is real.
But in public markets where things rarely go to zero, Kelly is reductive.
An investment that can return 200% at a -50% risk has the same output as one that can go up 400% but -10%
But in public markets where things rarely go to zero, Kelly is reductive.
An investment that can return 200% at a -50% risk has the same output as one that can go up 400% but -10%
10/ My current personal framework is
(p/a - (1-p)/b) * x
Where p = prob of win, a = magnitude of loss, b = magnitude of win, x = top down risk cap
(p/a - (1-p)/b) * x
Where p = prob of win, a = magnitude of loss, b = magnitude of win, x = top down risk cap
11/ Benefit of the above is it highlights asymmetry easily (low a), but a pre-defined size limit (x) for a specific type of bet
e.g. Capping core holdings to 50% of bankroll, catalyst play to 25%, venture stage to 10% allow me to control risk
e.g. Capping core holdings to 50% of bankroll, catalyst play to 25%, venture stage to 10% allow me to control risk
12/ It's not perfect but at least it gives me a barometer to gauge my decisions.
Does it work? Guess we'll find out in 10 years...
But point is having a systematized way to inform discretionary bet sizing is pretty key.
Does it work? Guess we'll find out in 10 years...
But point is having a systematized way to inform discretionary bet sizing is pretty key.
13/ Finally - a common trait of successful investors in history is their propensity to place large, concentrated bets when asymmetry is identified.
These bets may occur once a year, or once in a career...
However, beware survivorship bias.
These bets may occur once a year, or once in a career...
However, beware survivorship bias.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
