Just 1 trade per year? If that is what you meant then no way I will be even trading anything based on that information. 5 backtest trades mean nothing.
Backtesting is useless
- Thread starter JCDST1979
- Start date
Just 1 trade per year? If that is what you meant then no way I will be even trading anything based on that information. 5 backtest trades mean nothing.
Well they would mean something if you tested a system that's supposed to produce a winning probability of say 98 - 99%. Perhaps HFT in some niche areas can reach that.
Either way I'm still wrapping my head around the gambler's fallacy and the Monte-Carlo casino situation: https://en.wikipedia.org/wiki/Gambler's_fallacy#Monte_Carlo_Casino
"Perhaps the most famous example of the gambler's fallacy occurred in a game of roulette at the Monte Carlo Casino on August 18, 1913, when the ball fell in black 26 times in a row."
Now even under the reasonable assumption of a fair toss (50% probability), getting one side 26 times in a row is not falling under "probability laws" any more, it's a freakin' black swan event.
What you can learn from such events:
- Never bet your castle, even on high probability events (>90%), let alone in the 50% range
- Apply a reasonable limit for "bad luck", say 1 in 10,000 events. For a fair coin (50% probability), that's some 17 tosses. Think if it as "not that unusual" not to get a win in 17 rounds, if the winning probability is 50%. So if you really are sure on that 50% probability, you gotta have at least capital for betting 20 times. But as the Monte Carlo Casino event shows, that's only a rough guideline, in practice you need to have a lot more - I'm taking a risk and saying that having runway for 100 bets ought to be enough.
- The problem with low probability bets is that it takes a long streak of loses to figure out you're wrong. So if you believe the win probability is 50%, you can still lose 20 times in a row and can't disprove that the win probability still is 50%. If you lose 50 times in a row though, then surely it's not 50%. Therefore, go as high as possible in the probability of your bets.
Thinking about it, I'm gonna open source my options backtester afterall. For the guys that don't know programming, they could still download the already built application and run it but I guess it wouldn't be of much use to backtest some precoded strategies - which as I pointed out earlier in this thread, don't work because everyone is using them.
Still you could see with your own eyes how:
- If you add enough spread (ex: bet on 50% proability with a 10% spread), the theory actually works in general
- The market is too efficient for "enough spread" when using the well-known theories
- When using the well-known theories and sometimes the market appearing to allow for "enough spread", say theory says 50% win probability and you can sell at 60% according to your model... well, theory is wrong, and win probability is actually 20-40%. And oddly, "the market" knows it, so the big banks probably hold more knowledge than you can get from published textbooks.
I've proposed this earlier, but maybe not very coherently after a few sixpacks, to get an edge you need to fall on something that "the market" hasn't yet thought of. That's not impossible even for a single person if they were Feynman, but since you compete with teams of Feynmans at the big banks, I think it's reasonable to try to increase your retalier odds by also forming closed research groups.
Discussing stuff on Elite Trader has it's value but once you're on to something, it would be better to keep it private.
Still you could see with your own eyes how:
- If you add enough spread (ex: bet on 50% proability with a 10% spread), the theory actually works in general
- The market is too efficient for "enough spread" when using the well-known theories
- When using the well-known theories and sometimes the market appearing to allow for "enough spread", say theory says 50% win probability and you can sell at 60% according to your model... well, theory is wrong, and win probability is actually 20-40%. And oddly, "the market" knows it, so the big banks probably hold more knowledge than you can get from published textbooks.
I've proposed this earlier, but maybe not very coherently after a few sixpacks, to get an edge you need to fall on something that "the market" hasn't yet thought of. That's not impossible even for a single person if they were Feynman, but since you compete with teams of Feynmans at the big banks, I think it's reasonable to try to increase your retalier odds by also forming closed research groups.
Discussing stuff on Elite Trader has it's value but once you're on to something, it would be better to keep it private.
Thank you. Very informative.
If 1 trade a week for the past 30 years, then yes I will consider it if there is still an edge after slippage. It's a totally different scenario from your previous hypothetical scenario of 5 trades for the past 5 years.
Look at the bolded part in the quoted messages below. I have underlined the key word just in case you misread it again.
While you are at it, make sure you did not misread your own backtest results.
