Modelling the Drawdowns to Improve Performance

R

Rapunzel

Hi Folks, I was just wondering if anyone has tried to identify and model market properties that may effect the maximum drawdown of an open position.

Assuming you trade with a fixed R:R in general, what is the maximum drawdown as a percentage of your risk, on the winning trades. And what is the distribution of the maximum drawdowns (as %) on multiple winning trades.

Is it possible to increase your profit by reducing your SL size whilst keeping you TP constant. Can you find an edge in the trade-off between the increased profit per trade, versus slightly lower win rate that results from reducing the SL.

I've had a quick look and I think there might be something in the property 'Directness': how 'direct' the price movement is, on a timeframe higher than the open position.

for a period of 14 bars, the 'directness' would:

Directness = (Close[0] - Close[14]) / (14 x ATR(14))

The average change in price over 14 bars, divided by the average true range in 14 bars. To measure the fractional change in price in a given direction versus the overall price volatility over the same period (using atr as a measure of volatility).

The Directness value ranges from -1 to 1 and is normally distributed. There is a higher tendency for the value to revert towards the normal in at the extremes. In some cases I found that if the value is high future values will be also, in other the opposite.

I think it would be probably best to look at the transition probability between different values of this property to predict where it will be in the near future, at a higher timeframe than your trade since it would be easier to predict nearer term results and the markets are fractal.

I am not sure exactly how to use it yet, any ideas?
 
Only two cents I have to add to discussion (you might think it is worth -2 cents lol) is to ask why would someone trade with a fixed R/R and as well as what is to be gained using that in your premise?
 
Yes.. on many occasions cut loss and having to see it go back up and past the stop loss and continue to rise. As for me, there is no hard and fast rule. My rule of thumb, a wider stop loss is necessary in a volatile market environment.
 
Peter8519 said:
Yes.. on many occasions cut loss and having to see it go back up and past the stop loss and continue to rise. As for me, there is no hard and fast rule. My rule of thumb, a wider stop loss is necessary in a volatile market environment.
More...
%%
SOUNDS good\a single stock could go to zero\goose egg. So cant really predict/that what stops are for. 14 candles could be part of a battle tested plan,50or 200 could also. NOT a prediction, not insured by any federal agency.
Something to be said for buy SPY every month/40 years/ a common way to be market millionaire.
 
SunTrader said:
Only two cents I have to add to discussion (you might think it is worth -2 cents lol) is to ask why would someone trade with a fixed R/R and as well as what is to be gained using that in your premise?
More...

Hi @SunTrader; thanks for the post. I currently trade 'moments' in a particular market, for example the moment of a breakout beyond 2 sigma on 50 bars in a currency pair. In that case the SL and TP would be a fixed percentage of the current sigma, say 1(tp) : 2(sl).

The process is to capture all those moments in a historic period of say 10 years and find properties in the market that I use to reliably tilt the probability of winning those trades from the break-even win rate (66.67%) to say 77.67%, in the model data and then on unseen future data.

The resulting signals have a suggested SL and TP in that ratio 1:2 in the H1 candle. Before executing the trade I would normally check the price action in a lower timeframe; M30, M15. And sometimes decide the to reduce the SL from its suggested point based on the price action in the lower timeframe.

The problem is that there is uncertainty in what would be the long term effect of those interventions on the SL, away from what has already been modelled to prove edge on 1:2 as recommended.

I want to model drawdowns so that I can continue to change the SL based on what I see in the price action with a level of certainty, that it will not negatively effect my profitability in the long run. Since reducing the SL distance whilst maintaining the TP distance will have an impact on the win rate.

If I change it from 1:2 to 1:1, I cannot expect the normal (no edge) win rate to remain at 66.67%.

If risk £2,000 to make £1,000 per trade for example on 1:2, then change the SL to risk £2,000 to make £2,000 on a 1:1, the profit per trade will increase, but I should also expect the win rate to reduce. I am trying to find a property that I can use to find the right balance in that trade-off: by how much can I reduce the SL and increase or retain the long term profit, over 300+ trades.

I think there might be an answer in the property mentioned above. But I have not tested it properly. The ultimate goal would be to know statistically, how much I can reduce the SL on pair, without decreasing long term profitability, so that I can confidently adjust the SL according to what I see in the price action on a lower timeframe in future.
 
Last edited:
ATR is a better approach IMO.

Ever watched any of former Goldman Sachs trader Anton Kreil's webinars?

Think this is one of his most recent (a bit long @ 2 hours) but he rightfully trashes R/R use.

 
SunTrader said:
ATR is a better approach IMO.

Ever watched any of former Goldman Sachs trader Anton Kreil's webinars?

Think this is one of his most recent (a bit long @ 2 hours) but he rightfully trashes R/R use.

More...

Seen. Where does it mention r:r?
 
Rapunzel said:
Seen. Where does it mention r:r?
More...
Sorry. Not going to watch it again to find part(s) but its there. He talks about the market does not care where someone's R/R is in price action. It is useful as a guide and nothing more.

How I feel as well.
 
Hmm...I dont mind watching it again but I'm pretty sure it doesn't say that anywhere. There was an interesting comparison between the approach of professional versus retail traders at the beginning. I think maybe you should watch it again.

.edit: oh I think maybe you mean when he said some retail traders focus on risk management and think that that alone can make them money when they don't have a good trading idea to begin with.

Yes, I also believe that you cannot generate edge with risk management alone. That does not mean you shouldn't have any risk management or r:r ratios. It doesn't really follow.
 
Last edited:
You must log in or register to reply here.
Back
Top