Video wird geladen...
Video konnte nicht geladen werden
"every single trader out there seems to make the same mistake over and over again. their position size is probably two to ten times more than it should be." martin shkreli built a kelly criterion simulator to test exactly this. "most folks don't actually have an edge when they... show more
341,984 Aufrufe • vor 2 Monaten •via X (Twitter)
33 Kommentare

This is called the Gambler's ruin. You can have an edge against the casino, but with large enough bet sizes you are almost guaranteed to lose it all due to random chance.

Interesting how this kind of content surfaces when its time to increase longs size

And people always make a big deal about how much cash Berkshire/Buffet always have. Good trades or investments don't come around that often. Sitting in a ton of cash equivalent is usually the right thing.

The issue wasn’t position size. Once all the shorts and longs are correlated (as part of the AI trade), all the small positions compound to one large oversized position

People still listen to him it’s actually peak stupidity

True. And the trap is even deeper - you cant size right before you validated the edge. Kelly needs your real win rate and payoff, and people like to plug in numbers from broken backtest or from imagination. Overbetting is just downstream of overestimating the edge.

Not sure why we would listen to Martin Shkreli. Probably better to talk to a Quant or someone who understands the (not very complicated) maths. The problem with this entire analysis is that to assess your likelihood of win you have to have a very clear idea of the probability of your win or your loss. Now you can do that in poker much like Edward Thorpe did but it's a damn lot harder to do with equities when the probability is a far less certain. And because the result is quite sensitive to these small changes in probability estimates you can come to what I call the "Kelly Cliff" which gives vastly different answers from not entering at all through to leveraging up. I did some modeling on this a year or two ago. I think it's still useful within the confines of recognizing its inherent problems with equities. But let's not to get too carried away

Gamblers ruin is an easier concept to understand and is related to Kelly. The house (or market) with infinite resources can withstand tail events you cannot

Traders are degenerate gamblers. When they're on the losing side of a trade, they think doubling down is the best bet. If they win, they get a bonus and think they are geniuses.

Kelly criterion suggests that you are not risking enough. For a system with a 2:1 reward-to-risk ratio and a 50% win rate, the full-Kelly is 25% of your account per trade. A losing trade reduces the account by 25%, while a winning trade increases it by 50%.

Bonsai

I fell victim to this so many times within the past ten years. I 100% agree this is the problem every time. Had I been disciplined to size just enough I would have 10xed my initial bag. But I always got greedy.

This is the formula if you have an estimate of your expected Sharpe ratio:

Wasn’t this the guy who went to jail for securities fraud and pumped up the price of some drug causing misery to millions? And you're gonna listen to him? Good luck

he is somehow humble and relatable. his comment would have been perfect had he said assuming 1:1. is TBPN a reliable source in general? I did watch before thanks to a good friend.

Sizing for me is #1

Sizing and risk are skills that typically develop after you’ve blown up once or twice…if you have the drive, mental fortitude, and learn to approach Mr. Market with humility

That does not compute. That is equal to saying if a whale trades x amount he will win but a shrimp trading that same amount will lose.

It is rather amazing some people go busto with a huge 60:40 edge. Albeit, they don't necessarily trust that they have the edge that experiments tell them they have -- which is a good thing.

@Coolbreez016

@MartinShkreli built a Kelly criterion simulator, co-founded a drug company, rewrote pharmaceutical pricing theory, and livestreamed his work with more transparency than any Wall Street exec alive. Then they put him in prison. A man this brilliant deserves a pardon — not a conviction. #PardonShkreli #TeamSHKRELI

Wow! Shkreli says something really smart! What's the world coming to?

Hate em, but so right CC: @jord_ant

Use ATR to size with a defined $ risk....

betting like you're the main character is a bold strategy.

Kelly fraction is: f = p-q f = 0.60-0.40 = 0.20 Plain English: Kelly says invest 20% of your total capital each time.

Where can I use the simulator?

You don’t really need a simulator to figure out what size you should put on. No one actually tracks their ideas / returns by idea No one actually confirms thesis No one cuts when they are loosing Simple 3 things improves drastically - then use 10% -20% of Kelly max

real life scammer is getting on @tbpn incredible work guys.

Never go full Kelly

Yup.

Position sizing is one of the few levers you fully control. You don’t control tomorrow’s market, but you do control whether one bad sequence ends your trading career.

Sizing discipline matters more than thesis selection in most cycles I've covered. A weaker thesis sized conservatively outperforms a strong one sized aggressively when the drawdown comes.

