SMARTY - Smart Martingale

[BlackOpzFX2]

Introducing my newest system SMARTY - The Smarter Loss-Controlled Martingale. 7% to 8% sequence failure cost instead of whole account failure of traditional Martingale. Only one open trade per currency (no active grid). Designed to Survive 9 levels (97% Success) or 10 levels (98% Success) of 2X Martingale. PROMO: 33% OFF - Pay Only $99 per Month Until 12/31/2024: (up to 3 months of savings) UGDR64335

By controlling the maximum loss the limited martingale performs as a decent ‘recovery’ system. Additionally, since this system only uses martingale sizing and not the multi-trade grid approach it limits the amount of leverage needed even at the largest trade sizes.

I am VERY risk averse and always design my systems so that I’m generally unconcerned about the trades and never worried about an unrecoverable loss (I like my sleep). I didnt activate ‘loss recovery’ until about mid August and you can see thats when the system profit began increasing.

Even without the martingale aspect the core system was approaching per-trade profitability. My final testing revealed two methods to achieve profitability. One was more frequent ‘real tick’ optimization (was once monthly advancing to bi-weekly) and the other was martingale recovery. After a bit of testing and realizing I could squeeze 9+ levels out of martingale at <10% max loss (instead of margin-call account closure) I selected martingale as the profit turn.

In actual practice it has worked exactly as backtested and in the current month every better than expected. Even though the system can absorb 9 or 10 losses at 8% or lower DD I’m using multiple mitigation methods than can reduce the impact even further.

This was tested with the GBPUSD sequence that closed profitably on 10/1 (trade 8 of 9). Until that trade the system was +7% for Sep24 and even if the system had completely crashed the impact would have been only 3% to 4% and the system would have still closed September with a profit!

For Example: This is the system CRASHING with a 12-streak loss and you can see that the impact is only 4.62%. This depends on the number of loss prevention features that activate so I always quote the max loss possible of 7% or 8%. With a little luck the crash streak itself can end at BE or even mildly profitable (but I’m happy just to see 8% reduced to 4%).

So knowing I have 9 or 10 trades to play with I optimize with the goal of the max losing streak being less than nine loss trades and the average loss streak being well within the safety margin. Below I’ll show the current optimizations I’ll be running for October. Surprisingly the Sep GBPUSD parameters are still the active for Oct and the USDJPY optimizations only just exceed the previous set. A good marker of the algo stability.

Lastly, after adding the martingale ‘recovery’ to a fairly solid system has reduced this algo to a risk-management and probability system. Its funny because knowing that the vast majority of trade sequences will work out to be profitable (90%+) or have minor downside impact is strangely calming and even most crashes are recoverable within the next trading month.

I continue to optimize but I think I’ve squeezed most of the best results out of the system. In future posts I’ll detail more about the system but basically it’s just a collection of VERY simple trading concepts and is actually more of an active trade-manager than anything else. More info to come…

6 Month Backtests - Current Parameters: Currently the LIVE algo win-rate of 40% is exactly in line with the expected backtest results. In the future I’ll detail more about my extensive backtesting protocol. MT5 is actually AMAZING for backtesting with actual per-tick price/spread data, cloud computing, Walk-Forward analysis and more but I’ll expand on that topic another day and why I want to ‘Python’ this system in the future. See ya in the next post…

---

---

---

[SeriousTrading]

I strongly believe that a trading system should recover from its drawdown by its own merit and not from some martingale.

However, this idea is interesting, and can be used when a profitable trading system reaches a certain drawdown, on paper.

And, to the best of my knowledge, no serious backtests have ever been conducted on that topic.

[BlackOpzFX2]

@SeriousTrading

I strongly believe that a trading system should recover from its drawdown by its own merit and not from some martingale.

Believe me I’m just as surprised by the testing as anyone but going strictly by the probabilities the math shows strong profitability. Of course there are many differences between my implementation and traditional martingale systems. Also attaching martingale sizing to a 40% win rate robot surviving 9 levels means I should only blowup 1 out of 100 times.

I’m projecting a 35% win rate which only has a 2% chance of blowup. Also the robot has a 1:2+ RR so its straight math. The backtested equity curves show 98% chance of success vs 2% blowup chance can result in impressive and stable profitable upside. In addition, the loss is limited on the failure side and can be mitigated to small loss, BE or profit.

Of course, I control how much I martingale and I have hard equity closes at 7% to 8% and can survive for 9 or 10 levels. A stinging but far from fatal drawdown. As I stated I feel very little stress watching it trade. Especially seeing the recovery potential vs the inevitable controlled max loss.

[SeriousTrading]

@BlackOpzFX2

2% blowup

Hi BlackOpzFX2,

So if I understand correctly the system will blow up twice after 100 completed trades, on average?
And when you say blowup, you mean a 100% drawdown (or higher if trading on margin)?
Very interesting post by the way,

[BlackOpzFX2]

Blowup 2X for 7% or 8% loss (or less) for every 98 successful martingale sequences. One sequence is 1 to 9 martingale trades that ends the sequence with a profitable net outcome.

[SeriousTrading]

I have studied every possible martingale known to mankind and yours is the most interesting I must admit.

A level 9 martingale (1-2-4-8-16-32-64-128-256) will create a 511 unit loss and yet your drawdown will not exceed 7 or 8% of trading capital, is that correct?

[BlackOpzFX2]

Yep, you’ve got it EXCEPT my martingale doesnt follow the normal progression and adjusts RR on the fly so my net RR is often 2X to 3X losses which allows me to employ strategies not workable with the progression you noted. My gains are normal Marty 2X but my losses are often much smaller in relation.

[SeriousTrading]

@BlackOpzFX2

my martingale doesnt follow the normal progression and adjusts RR on the fly so my net RR

Care to give us an example, without revealing your secret sauce of course LOL ?
Thanks.

I also noticed that your win rate is low (absolutely nothing wrong with that), so I assume you are using trend following techniques, known for their small win rate percentage.

[BlackOpzFX2]

Yep, its Trendy. Martingale covers for the 40% or less Win Rate since a completed sequence closes in profit even though the Win/Loss count could be 1 to 8. The progression is part of the ‘sauce’ even though the stars of the show are the dynamic closures that reduce losses vs net gains.

[SeriousTrading]

@BlackOpzFX2

a completed sequence closes in profit even though the Win/Loss count could be 1 to 8.

Shouldn’t a successful martingale (it ends before reaching the 9th level) automatically create a new equity high, by definition, or are you are strictly using them as a quick drawdown recover mechanism.?

Anyway nice concept you’ve got here, adjusting a martingale to the risk/reward ratio seems like a promising idea, please keep us posted.

[BlackOpzFX2]

For this system late closures wont always create a new High-Water mark. After i get so deep into a progression this system starts looking for the door. The closer to failure the more willing it will settle for small loss, BE or small profit. Its rare and unlikely for it to take a full 8% loss. Its usually mitigated to some degree. The goal is to make losses more easily ‘recoverable’.

[SeriousTrading]

@BlackOpzFX2

The goal is to make losses more easily ‘recoverable’.

Got you my friend, your goal is to speed up your recovery time with a risk-controlled adjustable martingale when the system enters a period of drawdown.

I just added your system to my Watch List, wishing you the best.

[BlackOpzFX2]

MAJOR UPDATE: For SMARTY - The Smarter Loss-Controlled Martingale. My system is a trend following system and only has one major weakness that almost every trend following system encounters. ‘Bias Reversal’. One of the few indicators I use is a standard 200 SMA. If the price is over it BUY, if under it SELL.

Any system using a huge SMA has the problem of continuing to trade even as the trend starts to move against you. These systems stubbornly continue to trade (and often waste money) until the price crosses the SMA and they are re-aligned with the price movement.

For the entire time I have been working on my robot I have tried just about everything I could think of to combat and mitigate the ‘red-arrow’ areas. I’ve tested almost every documented indicator and trend strength detection method and have never been satisfied with the solution. Until NOW!..

Before today, my system entry parameters would still trade for profit even though its an undesirable area. With up to 9/10 trades to survive the reversal the EA can typically weather the storm at BE or mitigated loss BUT if I had my way it wouldn’t trade those areas at all. As a programmer once you see problems like this it’s too easy to create a curve-fit solution that would fix the backtest but be unreliable in the future. I’ve rejected over 30+ ‘solutions’.

This was my very last ‘back of mind’ issue that I thought was never gonna be ‘cleanly’ solved but the fix turned out to be a SINGLE LINE of code (Eureka!). It’s funny. As the EA has neared completion my mind always shifts to the ‘next’ nagging issue to play whack-a-mole with. Now I don’t know what’s left. This might have been ‘IT’ for the majors. I’m sure there will be minor issues, but I think I’ve hit everything on my operational ‘wish-list’.

GOOD GRIEF!! - This has been an agonizing issue and truthfully no indicator really helped. Not a single one. They ALL lagged too much and/or negatively impacted profits even if they reduced bad trades. I had grand hopes for ADX or MACD but nope. All it took was a simple objective rule to solve the problem 100%. What a freakin’ relief!! - I’ll be celebrating for the rest of the day…

LAST NOTE: Traders are Degenerates. I don’t have the emotional control to manually trade and this fix PROVES it. Even though I don’t manually trade I didn’t realize how addicted I had become to checking my phone and seeing active trades. It didn’t hit me until I realized the time length of the green-box areas. The largest is 9-days and the smaller is 4-days.

When it first hit me that I might not see a trade for a WEEK! even though not trading a single-time in the green-box areas is EXACTLY what I want - I didnt like that reality at first. It flashed in my brain to edit the code to trade more until I slapped myself! That I even had that instinct is absolutely ridiculous.

No way I could NOT trade for an entire week as a manual trader. It would take me years to not imagine setups that dont exist. BIG PROPS to you manual traders out there that can stick to your trading plan even when nothing is happening for days. I dont think I could do it.

[SeriousTrading]

@BlackOpzFX2

I dont think I could do it.

Same here, that’s why I day-trade.

Waiting for a trading setup for days, weeks or months before pulling the trigger and then babysitting the position for months is way beyond my mental capabilities!

The market is making nice moves EVERY single day, so why not profit from these daily moves and make money, instead of waiting for weeks/months for something to happen?

Some traders can wait 9 months for a confirmed head and shoulders formation before pulling the trigger. For me it’s a total waste of time, because they can make plenty of money daytrading for 9 months, using the SAME exact trading techniques that they already use on the higher time frames.

[BlackOpzFX2]

@SeriousTrading I read a reply you made in another thread about how you like the ‘Risk of Ruin’ stat. When I read that I laughed. ‘Great Minds Think Alike’. I keep a myfxbook on the SMARTY - Smart Martingale source account so I can isolate my stats from the date of the completed algo (or any future milestone changes).

It shows a few things I had mentioned previously about the algo. You’ll notice that the drawdown is lower excluding that previous algo but I’m happy as long as its under 15% which is half the expected 30%. You’ll also notice that C2 gains are higher than the source. Yep! - I filter the trades I pass to C2. I reserve the right to change my mind even AFTER a new trade is placed! Works well so far.

I’m LOVING! my current Risk of Ruin. I’m wondering how long I can keep it below 1% just for fun but it wont reflect the worst case reality until it experiences a full mitigated blow-up. The code to reduce the loss can actually flip a blow-up into BE or small loss/profit so stay tuned for the first full test of that system but I have real-tick backtests I wont even post because they’re too optimistic (Even I’m like. ‘Yea, Right…’).

Also you can see the current month RR and both currencies are currently over 2.0. Normal martingales are 1:1 and in Sep GBPUSD was 1.03:1 (USDJPY 2.66:1). Its all dynamic and this months its squeezing more profit out of GBPUSD.

The actual ‘printed’ to account results are even better because of the filtering. As I’m writing this post both Sep/Oct are at 5% gain on C2. Its the 16th of Oct. so even if we maintain the 5% we’re at my algo expectation. 15 days to go and I expect the markets to start heating up again next week. Should be interesting…

[BlackOpzFX2]

SMARTY - Smart Martingale had a pretty good day today (lots of week to go!). About 4 levels into progression trades for USDJPY and GBPUSD. USDJPY actually utilized a new update I mentioned in another thread that I had yet to see in action. Before recently I’ve been fixated in minimizing drawdown and squeezing the most profit out of reasonable risk parameters. My original goal was 15% which is half the typical 30% expected drawdown. At this point I’m pretty satisfied with all of my defensive code but I always had a nagging ‘want’ from my robot. Strictly optional but nice if you can figure one out.

And what I’m talking about is ‘Jackpot’ code. Is there a way your system can be open to a ‘lucky’ winner without increasing risk? Until a few days ago I would have said ‘No’ but now I can proudly say ‘Yes’ and it seems to work as intended.

Looking at a last 6-month backtest I’m kinda shocked at how close it is to my C2 performance. I shouldnt be surprised because my backtesting process is excruciating and MT5 with real tick/spread data is really ‘WYSIWIG - Whats You See Is What You Get’. Before this week my RR was about 1:2 and with the ‘Jackpot’ code it now has a ceiling of 1:4. (Payout Ratio - Avg Win/Loss) - Its a ‘soft’ ceiling since you cant depend on it but just a couple could boost your avg RR.

USDJPY closed out with a profit of $377 and when GBPUSD closes (and shows) it should have similar or greater profit so about 1.5% on a ‘good’ day with pretty low downside risk. Previously it would have still been a good day but prob closer to 1% so I netted a ‘lucky’ 0.5%. I’ll take it for no additional risk. It also helps the martingale since lower progression trades (with smaller lotsizes) can have a bigger positive effect if they get ‘lucky’.

Still 10 days until EOM so anything can happen but even in the event of pullback I should still meet my net goal rate of 3% to 5%. In any event for now I’ll can keep my fingers crossed and hope GBPUSD gets ‘lucky’ too.

[BlackOpzFX2]

SMARTY - Smart Martingale is also being used to drive my Darwinex-Zero account. Darwinex is an investment service that allows you to attract capital depending on risk-adjusted performance. Some find the service difficult to trade but it seems to like my robot. My current performance makes getting another allocation easier In December. Today I paused trading once I crossed the 75.00 mark to lock in my first 30K award since allocations aren’t awarded until the first day of next month.

My C2 account will continue trading the system normally so I’ll get to see ‘What If’ for the net monthly result C2 vs Darwinex. Next month I’ll let the robot run until months end regardless since each allocation lasts for 3 months. Also, kinda proud that I’m running a MARTINGALE on Darwinex which is VERY risk-adverse and its rating my system as pretty ‘safe’. In fact the risk-engine actually boosts the size of my trades to present the engine with normalized 6.5% exposure risk.

[BlackOpzFX2]

SMARTY - Smart Martingale performed pretty good in Oct. and I expect the train to keep chugging down the tracks. I’ve spent the weekend creating the latest parameter optimizations needed to update the robot for November. Last month actually outperformed its backtest expectation. I’m aiming for 3% to 5% a month with reasonable drawdown. Last month nicely overshot the mark.

Here are the November parameter backtests I’ll be using. I’ll prob run another couple optimizations this week trying to improve results. After that period I let the settings run until months end.

---

[SeriousTrading]

@BlackOpzFX2

I’m aiming for 3% to 5% a month

That represents a 42.58% to 79.59% return per year, on a compound basis.

[SeanKelly11]

Interestingly, you are on Darwinex Zero. I have an account there as well.