4.89 Sharpe Ratio

[dominicp]

I've developed an algorithm which after 3 months of trading gives me this:
- Annualized Sharpe ratio of 4.89
- Total gain of 15.5% (not annualized) over the 3 month period
- Over 1000 trades using a very diverse set of securities
- Gain above include transaction fees, bid/ask spread, etc.
- No leverage (average cash reserve are at 5-20%)

Do you think 3 months is long enough to know if I'm onto something? This means about 62 trading sessions.

The law of large number would suggest this type of performance cannot be caused by luck based on the number of transactions. At the same time, we're only talking about 3 months.

I don't know what to do:
1. Wait for more "evidence"
2. Invest more money
3. Enable leverage
4. Package/sell the IP
5. Something else

PS: I'm a software engineer by training. I've worked in nuclear physics, Microsoft, and co-founded a successful technology company.

 

[MikeLawrence]

If this is real, and you're making money, then just keep making money and see what happens.

As far as whether or not this is luck, 3 months is not enough time. Also, I'm not sure in what sense you're invoking law of large numbers. It is sorta agreed that market returns are not normally distributed (apart from lower level undergrad textbooks pretty much everyone agrees they have some fatter tails), hence a larger negative shock can pop up and ruin your 3 months of data (4 now). So if you think your strategy found some mean return higher than the index or whatever, it probably didn't. It's a short time frame, you probably just haven't been hit with the other shoe yet. But even if it did, the alpha is supposed to decay so I don't know how long you've got. Point being, your returns won't look smooth and stick to the current average you found. That's not something I'm an expert in (though this is why we are all here) but either way just off of time alone this isn't enough to say you've struck gold. And that ignores more factors than I even know of that should probably go into this.

But heck, you're a nuclear physicist who worked at Microsoft. You know more than I do, and this is more than I've done. But if you had to ask an online forum, I'd guess this isn't it.

I'm assuming that since no one responded to this yet that everyone sorta thinks it is spam.

But if anyone wants to comment on my analysis, feel free.
I finally found a professor that occasionally talks about fat tails but the professor can't actually go into any of it much because Calc I is the only pre-req, so he just makes aside comments to me every once in awhile since he knows I like the math. Point is, I have had no input on this knowledge. I just read two of Taleb's books when I was bored in HighSchool. If anyone has a bad opinion of Taleb from Twitter, and thinks this reflects poorly on myself, know that I also read Derman and he said the same thing (mostly) but in milder terms. Also Mandelbrot, same deal. And it's kind of implied in Spitznnagel, but I'm tired and can't remember anything specific. Any takers? Anyone have a paper they can give me to read? I haven't actually searched for anything technical until recently because I tried reading Taleb's technical stuff back in highschool and realized I needed more analysis/mathematical experience before I could touch that area. At one point Taleb was on the master list, and I haven't figured out why he got removed.