Adrian Reid compares two ways to launch a trading system: start fresh with new signals only, which delays profits but limits early losses, or seed it with positions from backtesting, which boosts returns immediately but risks big reversals. He recommends stress-testing by running the backtest from different start dates to see if the initial drawdown is acceptable.
So we're talking about initial drawdown and when we start trading a system, what happens and how do we make a choice about what we do when we're starting a new system? So when, when you've done the whole system design piece and make the decision to go live, there's two ways to do it. The first and most common way is to take every trade that is triggered from now on.
And what that does is it ensures that your initial risk for each trade is contained.
Because, say, for instance, you initially risk, you know, 2 ATR, which you equate to say, 1% of your total account on each trade. But as a trade grows, if you use a wide trailing stop, for instance, the instantaneous risk on each individual share could be quite wide, right? So most people start by taking new signals so that the, the risk on each new trade that they put on is contained.
The trouble is you don't have any of the old trades in your portfolio that are already winners.
Because they were triggered before you actually started trading. So the second option, rather than starting and taking every new signal, the second option is to on day one, replicate the portfolio that you would be holding had you always traded that system. And if on day zero, just before you started trading, your theoretical portfolio was full on day one, when you start trading, you're going to take trades which would literally fill up your entire capital. The disadvantage is if that trade that the theoretical portfolio had held for three years now turns around on day one and starts falling, it's going to go a long way down before it hits its exit point. The flip side, or the benefit of that approach is that if that trade that's already a winner keeps going up, your portfolio keeps going up. So it's a trade off between.
Having the initial trade drawdown, which occurs because you're getting losing trades, losing trades, losing trades, waiting for a winner to come through and drag your portfolio up because that's, that's difficult to cope with, as we found out. But eventually you make money. You just know that each loss is going to be easy because it's small.
The trade off is you don't have the initial trade drawdown. You actually get to take a sampling out of your historical long term equity curve from the backtest, and that's more likely what you'll get. But you may have a trade that has a big dollar loss. So it's painful. If you wanted to replicate the portfolio that would have been open, you just start your backtest 20 years ago forward and say you started with $100,000 you ran it forward and on the day you were going to start trading that system, the equity curve was at 3 million. Assuming you've got the same position sizing model through the entire backtest, take all of the positions that are in that $3 million portfolio and scale them to the amount of capital you're allocating to that system today. So then the position sizing becomes reflective of what it would have been had you always traded that system. But you know, it's a lot of execution on one day. You get a lot of, a lot of risk on each individual position. And if the market turns down on the day you start trading or soon afterwards, then you end up turning down having no real built up profit because it's all starting capital. So if you were going to do that, you want to be really comfortable with the worst case drawdown that you got and design your rules, your position sizing and portfolio management with that in mind, because that is the drawdown that you risking getting initially.
Compare it to what you have done, which is starting taking new signals on the day you go live. In that case, you've got to be comfortable with the initial drawdown.
And how do you get comfortable with the initial drawdown? You can't just run a 20 year backtest. You've got to start the back test at many different times and see how big the initial drawdown can be. You might start at 20, 30, 40 different points in time. In most automated trading software you could simulate that or replicate that by having optimization variable that says start date, trade trade entry date must be greater than X. And then step x forward, say one month at a time for 20 years and then at the initial drawdown. So what are the steps? Just take a big step back. You're gonna have your system rules done, you're gonna do the optimization, stability test, all of that. Then you're gonna say, hey, I'm about to start trading this. I better make sure I'm comfortable with the initial trade drawdown. And I better make sure my position sizing is gonna keep me comfortable because I don't want to freak out before I've kind of launched into profit. Do the step of the start date with your current position sizing rules. Analyze the initial drawdown, look at the range of initial drawdowns. Extract that automatically from most trading software by looking at the lowest equity during the backtest. If you, if you're uncomfortable with what that max initial drawdown is, go and change your position sizing, change your number of trades, or go and change your exposure cap or your leverage level. The chances are initial drawdown is usually going to be worse than the the well equal to or greater than the worst historical drawdown in a long term backtest. Because a long term backtest you've got the benefit of winning trades from previous years coming in. The ongoing drawdown is going to be lower because you've got winning trades bulking up the portfolio.
So if you really don't like the initial drawdown, but the long term backtest is amazing, what are you gonna do? You probably adopt the other option for kicking off the system.
Which is starting to trade as if you had always been trading that system.
