Is there anyone here who has a consistently profitable automated trading system?

[Laissez Faire]

I think a 90% win rate is exceptional and I almost wonder why the PnL doesn't go parabolic?

Accomplishing a 90 % win rate is easy. Just use a 50 point stop and a 1 tick profit target.

A 90 % win rate with a positive R/R - now that would be very impressive indeed and I'm sure only very, very few if anyone can do that consistently.

If your R/R is 1:2 you can still do well with a 50 % win rate as your expectancy is $50 bucks gross per trade assuming $100 risked per trade.

If you can do 1:5 you'll do very well with an expectancy of $200 bucks per trade assuming $100 risked per trade. In fact your win rate can drop way below 50 if you maintain that R/R and you'll still do okay. But I'm sure you already know this.

A lot of would-be traders don't, though. And get caught up in win rates instead of expectancy which effectively decides if you make money at the end of the day or not. I think Quallamaggie, Swedish swing trader, said his win rate was around 30 % last time he checked (he said something about not keeping track of that metric).

R/R 1:2 50 % Win Rate (100 trades)

R/R 1:5 50 % Win Rate (100 trades)

 

[NoahA]

Simple, the trades I posted was an account that only trades one contract at a time. That $19K day you saw is for an account that trades a maximum of 23 contracts with I think close to $700K in it currently. Big difference.

Also, the "one contract" account is our most conservative PnL simply because, trading one contract, there's no possibility to scale into and out of positions. Trading a larger number of contracts allows our system to use tiered exit targets from conservative (what you see in the one contract account) to aggressive. One entry can have 3-5 profit target exits if the trade quantity allows it.

Lastly, that $19K day, around 3% ROI (also in that screenshot), is obviously not an average day, or the monthly ROI I'd be reporting would be closer to 50-60%, not 5-15%...

Thanks for the explanation. I'm happy to see it isn't a totally different algo, but just a combination of multiple contracts and various exit strategies.

I think where most traders fail is that they find something that works, but then try and make it better, or do different things with it in order to improve some metric like a longer hold time. I always thought the easiest approach was simply to scale up. If you can make $50 per day trading a micro contract, you can make $500 trading the full contract, or $5k trading 10 full contracts. The psychological impact of this is of course huge, and that is I think where most people fail. But absent some fill anomaly, the market can likely absorb whatever us retail traders want to trade.

 

[NoahA]

Accomplishing a 90 % win rate is easy. Just use a 50 point stop and a 1 tick profit target.

Honestly, I don't think I could hit this. I have a skill for waiting until the move is practically done and then jumping on board. I could literally lose 50 points several times in a row since I'm really good at timing when my peak FOMO hits, and then I think I have to jump in since I can't take watching it move without me any more.

But expectancy is absolutely a critical metric. I figure most people aren't thinking at the extremes though. We generally do realize that anything below a 20% win rate would be a terrible system because even if a win would be 1000% and hence profitable, psychologically its very hard. Likewise, a 90% winning system would have most people be too scared to pull the trigger for fear of getting a loser, even if the metrics supported a positive expectancy.

What we all would like I think is something like a 60-80% win rate, so we feel good most of the time, and a better than 1:2 risk to reward so that if we lose the first time, we make it back and go into profit on the second trade. (of course you can still have 3 losses in a row easily with a 60-80% win rate).

Psychologically, losing 5 times in a row is very hard, but to be expected with a 50% win rate, and if you risk to reward is 1:2, you're actually golden.

As you can see, I do kind of like the win rate metric, but it of course has to be tied to expectancy.

 

[Laissez Faire]

Psychologically, losing 5 times in a row is very hard, but to be expected with a 50% win rate, and if you risk to reward is 1:2, you're actually golden.

Wouldn't that depend entirely on the stop size/risk? Big difference between 5 medium size losses and 5 small losses, IMO.

With day trading, it's generally hard to achieve those higher R/R ratios, though. It becomes easier with swing trading as the moves generally are much larger and you can still enter near extremes.

I won't disagree that it's a good aim to increase your win rate above 50 %, but my main comment was that I wouldn't consider a 90 % win rate exceptional unless I know the average risk/reward. A strongly inverse R/R ratio is typically a disaster waiting to happen. It can work well for a time, but eventually that bigger loss that takes away all prior profits comes.

 

[TrAndy2022]

The only thing to consider here is the overall Profit Factor which should be above 2.4. Because noble prize winners has estimated that a loss of $100 equals $240 profits on ones his emotions, that is because humans are usually risk averse. You cannot loose more than 100% but you can make multiple times more on gains. So this is understandable. Anyway your profit factor is above 1 but below 2.4 you should consider improving it, because if you trade this hardly profitable strategy you get emotionally negative over time which also can cause more trading errors.

 

[chris.page]

Thanks for the explanation. I'm happy to see it isn't a totally different algo, but just a combination of multiple contracts and various exit strategies.

Here's an example using today's trades in that account:

As you can see, one strategy opened a long of 17 contracts (in two transactions) at 07:01, and another strategy opened up a long of 6 contracts (in three transactions) at 10:02. However, the 6 contracts had three different exit targets, and the 17 contracts had two different exit targets.

The "conservative" account that only trades one contract for each strategy only saw the most conservative two trades above. Still winning trades, but therefore a lower ROI.

And it was another good day for that account (and the others running the same thing at various scales), five trades, five wins, up over 2% for the day on our beginning-of-quarter trade balance of $575K for that account, and at around $710K currently, up over 23% since July 1st.

 

[NoahA]

Wouldn't that depend entirely on the stop size/risk? Big difference between 5 medium size losses and 5 small losses, IMO.

I'm actually just using the data that Chris has provided. In that conservative model, there were only 2 losses for -875 and -583. Now granted, the wins are about half that or less, but with such a high win rate, its really no problem. In fact, if I run a Monte Carlo simulation on this, using a Risk:Reward of 3:1 with a 90% win rate, we get amazing results.

I won't disagree that it's a good aim to increase your win rate above 50 %, but my main comment was that I wouldn't consider a 90 % win rate exceptional unless I know the average risk/reward. A strongly inverse R/R ratio is typically a disaster waiting to happen. It can work well for a time, but eventually that bigger loss that takes away all prior profits comes.

I totally agree, but when dealing with automated trading, you've got a huge advantage. The algo won't feel the pain, and it also won't suffer the depression of having a bigger loss. Clearly the stats for the trades Chris has shown have an amazing expectancy. A human trader maybe won't be able to perform at this level, but the algo certainly doesn't have a problem. And the fact that the algo is always updating itself is literally the holy grail because it will adapt to changes that a guy who programs his own algo has to do manually. He will only make changes once he notices a drop in performance, but the algo is perhaps making adjustments in real time.

Also, the medium sized losses vs. small losses discussion does already set us up for failure. None of this language really exists with automated trading. The trade size is literally just a reflection of account size and trade expectancy. For you and me, medium vs. small loss has meaning of course. And if we consider that we took many medium losses vs. many small losses, that is already going to affect our next trade because of the introduced fear. But the algo wouldn't have this idea at all. It doesn't feel the difference between a small vs. medium loss because it's stop is based on the likelihood of the win. The medium loss it would be willing to take since the payout would be big, and the small loss is would take on a trade that perhaps isn't as likely. But my point is that there is no difference to an algo between medium and small loss because both of these are mathematically the same with reference to expectancy.

I think of it like this. I need to trade 10 km. I can either walk at 5km/hr and get there in 2 hours, or I can run at 10km/hr and get there in 1 hour. The running will leave me tired, so its more draining, but the walking will take longer. Both cause an emotional reaction in me and I have to balance my energy level with my time. The algo, if it was an electric car, just looks at its level of charge and figures out what the most efficient speed is taking into account how much juice is in the battery.

What's really interesting about Chris's trades if you plot them out is that they aren't based on stellar entries. The DD is actually sometimes bigger than the win. The entries aren't precise, although some exits are 1 tick from the end of the move, but that could just be a random distribution (even I can get a killer entry... its just that I exit after 2 points when it goes instantly in my favor... LOL). The exits also half the time miss a good chunk of the move. But as Chris said, banking those profits is generally better overall than going for the home run. If you get a +20 points trade twice, but miss out on +7 points for many other trades, you're actually not better off overall.

It seems to me that the algo enters kind of late because it waits for many signals to line up (as he said with the 66% confluence metric). At this point, the entry is no longer precise, and you don't know if it will still retrace, which it often does, but the win rate is much higher than if entering at the beginning of the move before you know if it is in fact trending or not. So the algo has the stats on its side to be confident to hold for a profit since its likely, no matter how long it takes to get there. I don't have the stats on my side in my head like the algo does, but I also wouldn't have the the mental fortitude to hold a trade going against me if I enter late. So the algo has me beat by a long shot.

 

[NoahA]

As you can see, one strategy opened a long of 17 contracts (in two transactions) at 07:01,

It's hard to fault the strategy given the stellar results, but man, the DD on this trade was pretty severe. The win is smaller than the DD, and as I mentioned above in a different reply, I have seen this in many for your trades. Most people would say this is a mortal sin, but you're obviously making it work.

So I guess the important question is why is the algo feeling bullish to enter there. We have all seen how easy the market gets pushed in one direction, only to finish by fully retracing the push. So is there some indication of this in the price action? Do you know why the algo enters where it does and why it doesn't exit for a loss and continues to hold? Is it simply that some oscillator or moving average on a higher time frame chart is keeping you in the trade?

I think this might be similar to some of the stats that @Laissez Faire collects where he gets a high probability for a market finishing above or below a certain value. If before the open the algo sees a close above 4500, so you enter at 4492, but the market tanks, this I guess doesn't invalidate the trade, but it sure does make a manual trader feel uncomfortable.

For those curious, here are the trades marked on a chart. (green dots are long entries, black dots are exits)

 

[chris.page]

It's hard to fault the strategy given the stellar results, but man, the DD on this trade was pretty severe.

Ha, "severe" is a subjective term. On a $575K trade balance, I don't consider an $18K MaxDD (thanks China) "severe". Not preferable of course, but no drawdown is.

So I guess the important question is why is the algo feeling bullish to enter there. We have all seen how easy the market gets pushed in one direction, only to finish by fully retracing the push. So is there some indication of this in the price action? Do you know why the algo enters where it does and why it doesn't exit for a loss and continues to hold? Is it simply that some oscillator or moving average on a higher time frame chart is keeping you in the trade?

Sorry, those are "secret sauce" questions .

this I guess doesn't invalidate the trade, but it sure does make a manual trader feel uncomfortable

For sure. Good thing I'm not a manual trader.

 

[chris.page]

As you can see, one strategy opened a long of 17 contracts (in two transactions) at 07:01, and another strategy opened up a long of 6 contracts (in three transactions) at 10:02. However, the 6 contracts had three different exit targets, and the 17 contracts had two different exit targets.

Also, what may not have been obvious at first glance is how these two strategies, while semi-independent, complimented each other. One strategy went long with a 17 contract position, but then a drawdown. That strat held through the DD, AND the second strat entered a smaller long at nearly the bottom of the DD, thereby reducing the overall average entry price (a pseudo dollar cost averaging play). If the first strat ended up stopping out for a loss, the overall loss for the day would have been less (unless of course the market kept tanking after the second strat entered near the bottom, which obviously it didn't, and is not likely when multiple strats confirm opening a position in the same direction).

Point being, one of the factors we take into consideration, in real time as well as tuning, is what strats compliment each other the best given their current circumstance. In this instance, it was a DCA play (and not just coincidence). In other instances, one strat being in a certain position can be the confirmation another strat needs to enter a position. The strats we run are more interdependent than fully independent.

There's your free peek behind the curtain for today .