Episode 974: Martin Shkreli Reveals Why Small Bets Make You A Better Trader

**Martin Shkreli** (0:02)
Yeah, Kelly's pretty simple.
So the basic idea behind Kelly was, there was a guy at Bell Labs, actually came up with the term member of technical staff back then. That's where it comes from, Bell Labs. There's a guy named Kelly, and Kelly came up with the Kelly Criterion, which is a proof. It's not like a heuristic, it's proof that the most optimal bet size for any given bet is actually your edge minus the reciprocal of your edge. So if you have a 60% edge, an edge means an expected value. For example, you have a coin that flips 60% heads and 40% tails and pays you a dollar or you lose a dollar, that's a 60% edge. That edge is the right edge amount is 60 minus 40, which is 20%. So it's the optimal bet size. So even with the 60, 40 edge, which is momentous, if you bet half your stack on this, you quickly go to zero. You could have a big run, of course, because you have a big edge, but generally, you're going to lose it all. And part of that is just do metric compounding returns. So you shouldn't bet 50% unless you have a much bigger edge, like 75, 25 That would be more normal bet amount. You could see this 10,000 ran to a million, but back down to $10. So sometimes you can go on a crazy run, but this guy's now broke. So it made 574 trades or coin flips and ended up with a bankroll of zero. So you may wonder if this feels familiar. If some of your trading is like this, then you should take a step back and think about, okay, well, what happens if I do the Kelly optimal? 20.2% in this case. Yeah, there's some Thorpe here. Thorpe worked with Kelly, if I'm not mistaken, for example.
And you could see the 10,000 becomes 10 million pretty quickly there. Now, who has a 60-40 edge? Not many people. So you have to be realistic. Every single trade you have has a 60-40 edge. That's, or on average, your trades have a 60-40 edge. That's very uncommon. Yes, if your sizing's, if your edge is 52% or 51%, that's still a pretty good trader. So it shows you that if you're patient and you select good trades, you can actually do really, really well. And I've seen this happen with many traders. But if you're impatient and you're trying to oversize your bets, you will end up failing. And that, you know, what's funny is you can still be a really good predictor. But at that point, you're not a really good trader because, you know, you don't have the discipline and the risk, you know, controls to avoid having a huge mistake.
Intel's 100 now.
Yeah, semiconductors are being bit up. Micron's almost 900 again. This is after Armageddon, you know, three, just three, four days ago. The market's pretty schizophrenic here. We'll see, we'll have to see what happens. But yeah, it's all right. We'll figure it out.
So let's see, even with the 70% edge, what happens when you double down your bet?
Is that a good idea? You double down when you're wrong. Is that a good idea?
The problem with doubling down when you're wrong is you'll be wrong sometimes, several times in a row, and you see how the drawdowns affect you. That's a very big drawdown model. Now, it's 70% edge, so that's quite a lot of edge. But you can see even with 70% edge, this poor bastard ran it up 10K, all the way up to what is that, 50 mil, and still ended up at zero, because of that Martin galing. Now, what about reverse Martin galing? That's when you get more conservative every time you win. Now, it looks pretty bad too, even with the 70% edge. I mean, 70 is big. Nobody's right 70% of the time. Maybe Buffett.
So you see how hard it is to get it right. Some people invest more as they lose. That's wrong. Some people invest less as they lose. That's wrong too. The right thing to do is make sure you have any kind of edge at all, even if it's 55%. Identify the kind of trades that work for you. And then do them consistently the same way. Don't get flipped out by the market and make sure your size isn't too high. Martingale does not work if you have infinite capital. Let me do a 1% size bet here, and a 51% edge, and try Martingale. Yeah, this is what I made this tool for. For comments like that. Let's see how that goes for this guy. It's not looking good. Yeah, consistent, consistent, but look at those dips. All those dips is going to kill. Now, to be fair, with 1% base bets, it would take a lot for this guy to lose, right? He's got 51% edge.