Road to $1M: Finding Lucky Trades

There are at least one thousand new lucky trades available in the market every minute. They are hard to find. I can’t teach you how to find them all, but I can teach you a process to find lots of them.

A lucky trade is a trade that outperforms your best known strategy. It is lucky because there is no element of your will or agency which explains the high performance of the trade. It is the best you could expect to do plus some probability of a random windfall or bonus on top. That’s why the trade is called lucky.

There are three reproducible ways to find a lucky trade that I can teach you. There are certainly other ways to get lucky, but none that I’m currently able to teach you in a reproducible way.

The first is to find a new strategy and enter the strategy in a lucky position. Your strategy will find some ordinary entry position after you deploy it. You can find lucky trades by considering the time before you deployed as well.

The smarter thing to do is to deploy the new algorithm and run a lookback or replay routine. Replay the algorithm over the last relevant period. You will need to assess what the relevant period is arbitrarily based on the kind of algorithm (daily trades or hourly? regime-sensitive or not?) and availability of data. If you’re not sure, looking back one month is a good and fairly generic default.

Share

When you run this lookback, have your program ask the following questions:

1. At every algorithmic interval, would I have bot some assets? What prices and quantities?

2. Would I have sold some assets? What prices and quantities?

3. Playing forward until the present day, would I be holding any assets? What prices and quantities?

Now you can check and see if you were lucky. If you were lucky, you can purchase the algorithmically desired assets today at price approximately equal to or even less than your high performance algorithm would have done.

So, you can find lucky trades through this lookback protocol on a new algorithm. You can also use a proven legacy algorithm! A proven and deployed algorithm would have already entered the position, so how can we generate lookback luck? By adding capital. If you have capital to allocate, you can add it with a lookback and potentially increase an existing position at a discount.

The third lucky trade is a lucky exit, and it exploits irregular intervals with arbitrary checks. Algorithms typically enter and exit on an interval, such as once per hour, day, or even each minute. If your algorithm would exit at the end of the day, you can potentially arbitrarily exit early for larger than expected return by checking during the day. There’s risk with these early exits: You need to be able to project the expected end-of-window return or price with confidence. If you can’t do that, you aren’t making a lucky exit. You are just adding risk. If your particular algorithm or the related assets can be projected on price or return with confidence, though, you might find that closing out the trade now is even more lucrative! That would be lucky.

This same lookback protocol is fantastic for crash recovery and system health in your automated trading programs. If your program crashes and you would have missed a trade, you can often lookback and recover it, sometimes at a discount!