[Darwin] UKC by Morpheus33

[AriaS]

Hi everyone, UKC is up . I was sure that no good names were left by now. But this one is like UK + See. Or is it a sea?

I wish you all an entire sea of prosperity!

 

[AriaS]

Update: I constantly keep working on the strategy and improving it.
The main latest development is 2 new back testing algorithms. One is testing 3 months back and the other one also takes into account around 3 problematic time segments in the past.
Need 3.2% this month for the first 30k allocation

 

[AriaS]

Random win probability

I often like to calculate the random win probability of my strategies to see whether it's luck or an edge:

First, we need to calculate the probability of a trade to become a winner.
Here's the formula: (average loss - average spread) / (avrg loss + avrg win)
My strategy: (42.45 - 1.5) / (42.45 + 22.3) = 0.632

This means that a trade has a random probability of 63.2% to become a winner. That's very reasonable: an average winning trade of my strategy is smaller than an average losing trade, so a TP has a higher probability to happen than an SL

Now we must calculate the binomial distribution. I like this online calculator .
I have 71 winning trades out of 84, so "number of flips" = 84, and we need to have "at least" 71 heads, at "probability of heads" = 0.632.
The result is 0.000014375 or 1 in 69,566 chance of success.

This indicates that the outcome is a result of an edge, rather than luck.

Why is this calculation important? For example, if I had only 60 winners out of 84 trades, the strategy would still be in profit, but then the probability of that to happen would be 1 in 14 . Such high probability could suggest luck, rather than an edge. We would not be confident in such a strategy's success in the future.

 

[Rufus_Leakey]

Random win probability

I often like to calculate the random win probability of my strategies to see whether it's luck or an edge:

First, we need to calculate the probability of a trade to become a winner.
Here's the formula: (average loss - average spread) / (avrg loss + avrg win)
My strategy: (42.45 - 1.5) / (42.45 + 22.3) = 0.632

This means that a trade has a random probability of 63.2% to become a winner.

Shouldn't the spread be added to the numerator instead of subtracted?

https://www.perplexity.ai/search/given-average-win-average-loss-L9NeJhp1QSae3v0zguKMLA

 

[AriaS]

Shouldn't the spread be added to the numerator instead of subtracted?

https://www.perplexity.ai/search/given-average-win-average-loss-L9NeJhp1QSae3v0zguKMLA

View attachment 342832

It should be subtracted, and here is why: spread shifts our entry against us, therefore TP is harder to reach. The probability to reach it with spread=0 should be higher, than with spread>0.

Let's check two examples that both have SL=TP=10 pips. The 1st has no spread and the 2nd has spread=1.

1. Spread=0: (10-0)/(10+10)=0.5 : The probability to reach TP is 50%

2. Spread=1: (10-1)/(10+10)=0.45 : Here, when we do have spread, the probability to reach TP is lower - 45%

 

[Rufus_Leakey]

It should be subtracted, and here is why: spread shifts our entry against us, therefore TP is harder to reach. The probability to reach it with spread=0 should be higher, than with spread>0.

Let's check two examples that both have SL=TP=10 pips. The 1st has no spread and the 2nd has spread=1.

1. Spread=0: (10-0)/(10+10)=0.5 : The probability to reach TP is 50%

2. Spread=1: (10-1)/(10+10)=0.45 : Here, when we do have spread, the probability to reach TP is lower - 45%

According to the link I posted above, the calculated value is the break-even win rate.

So, for SL=TP=10 pips comparing zero spread and spread=1,

1. Spread=0: (10+0)/(10+10)=0.5 : The break-even win rate is 50%.
   For example, 20 trades with 10 wins and 10 losses (50% win rate) breaks even ((10 * 10) - (10 * 10) = 0).
2. Spread=1: (10+1)/(10+10)=0.55 : With the spread included, the break-even win rate is a higher 55%.
   For example, 20 trades with 11 wins, 9 losses, and 20 spreads (55% win rate) breaks even ((11 * 10) - (9 * 10) - 20 = 0).

It's not possible to calculate the probability of reaching a target point given only average win, average loss, and average trade cost.

https://www.perplexity.ai/search/given-average-win-average-loss-rix1TMdKT6iwv3vT1OqE4w

**Key caveat:** You cannot calculate your *empirical win probability*—the actual chance that any single trade is a winner—without knowing either the number of winning trades or the win rate from past data[3][4][5]. You can only determine the *break-even* win rate using averages[6].
...
[3] https://www.quantifiedstrategies.com/trading-probability/
[4] https://www.quantifiedstrategies.com/win-rate-trading/
[5] https://www.luxalgo.com/blog/win-rate-and-riskreward-connection-explained/
[6] https://market-bulls.com/breakeven-win-rate-calculator/

 

[AriaS]

According to the link I posted above, the calculated value is the break-even win rate.

So, for SL=TP=10 pips comparing zero spread and spread=1,

1. Spread=0: (10+0)/(10+10)=0.5 : The break-even win rate is 50%.
   For example, 20 trades with 10 wins and 10 losses (50% win rate) breaks even ((10 * 10) - (10 * 10) = 0).
2. Spread=1: (10+1)/(10+10)=0.55 : With the spread included, the break-even win rate is a higher 55%.
   For example, 20 trades with 11 wins, 9 losses, and 20 spreads (55% win rate) breaks even ((11 * 10) - (9 * 10) - 20 = 0).

It's not possible to calculate the probability of reaching a target point given only average win, average loss, and average trade cost.

https://www.perplexity.ai/search/given-average-win-average-loss-rix1TMdKT6iwv3vT1OqE4w

Please, paste my previous reply to the language model that you are using. We are not looking for a break-even rate of the entire strategy. We are looking for the probability of any given random trade to become a winning trade (you can tell this to the language model, and it will understand its mistake).

 

[Rufus_Leakey]

Please, paste my previous reply to the language model that you are using. We are not looking for a break-even rate of the entire strategy. We are looking for the probability of any given random trade to become a winning trade (you can tell this to the language model, and it will understand its mistake).

You are correct for a random walk model if all trades have the same target, stop, and spread.

https://www.perplexity.ai/search/is-the-premise-presented-below-ZbTLpgYWQ8yYLv.rwXkT4g

Summary: Using only averages, you can estimate expectancy, not the underlying probability of each trade winning—unless your trading strictly follows the random walk model with fixed TP, SL, and cost for all trades. Most trading reality is more complex.

 

[AriaS]

You are correct for a random walk model if all trades have the same target, stop, and spread.

https://www.perplexity.ai/search/is-the-premise-presented-below-ZbTLpgYWQ8yYLv.rwXkT4g

I disagree. We can use averages for the purpose of this calculation. Of course, it will be approximation, same as all the trading always is. Btw, ChatGPT agreed 😉

 

[AriaS]

Update - 1st allocation & prop trading:

Received the 1st allocation: 30k. Going on vacation from trading until Aug 1st. In the next month will need to make only 1% for another 30k.
Still, I think the 15% performance fees on DarwinIA is too low, so I am recently considering to try prop trading. For the 1st time I see that I have the chance to make it.
I found a really nice and cheap offer from a large and trusted company. Not sure if I am allowed to share here.
It has one annoying restriction - no trading at all during the news - but my strategy rarely does it anyway (spread and slippage filters).
I've already coded a new time filter for it. The generous offer is $180 for 50k account and a profit:loss ratio of 1:1. Would you try it?

 

[IlIlIlIlI]

I found a really nice and cheap offer from a large and trusted company. Not sure if I am allowed to share here.

For sure you can publish the name of the company, FTMO and other prop firms appear in tons of posts here. Maybe you should not post a link.

It has one annoying restriction - no trading at all during the news - but my strategy rarely does it anyway (spread and slippage filters).

I'd recommend to use a demo first.
If your finding does not have a demo, FTMO has (for registered users). Unfortunately you have to check the restrictions against your trades manually as the demo does not tell you your violations - the live system will do.

If you're trading with sl and tp, you can always be triggered during news releases. Some prop firms also don't allow open positions in the related currencies during news releases.
If you're trading overnight, you should also check the T&C whether that is a rule violation.

You will learn the rest of the traps on practice.

 

[AriaS]

For sure you can publish the name of the company, FTMO and other prop firms appear in tons of posts here. Maybe you should not post a link.

It was Oanda Prop Trader, but I didn't take it. I was really disappointed. The conditions were good but I simply couldn't pay them: malfunction on their side and they kept saying it's my credit card, while both my credit card company and the bank refuted it. I had the feeling that OPT simply didn't want me to buy the challenge. It had taken their customer support ages to reply. They probably had enough buyers at that discount.

If you're trading with sl and tp, you can always be triggered during news releases. Some prop firms also don't allow open positions in the related currencies during news releases.
If you're trading overnight, you should also check the T&C whether that is a rule violation.

I don't use hard stops. My EA uses market orders. Yes, I've deeply learned the rules of at least 10 firms and came to conclusion that they will probably always find a reason to disqualify. I don't fully trust them. I think their only source of income is challenge fees and profitable traders are simply an unwanted byproduct. Because in order to offset the payments to them, the firms need to start copying them on live accounts at some point. But that's a risky business. Why would they engage in it if they have such a nice flow of risk free money as the challenge fees. But I don't know. Maybe I still should try at least a small account at FTMO. What do you think?

 

[IlIlIlIlI]

It was Oanda Prop Trader, but I didn't take it. I was really disappointed. The conditions were good but I simply couldn't pay them: malfunction on their side and they kept saying it's my credit card, while both my credit card company and the bank refuted it. I had the feeling that OPT simply didn't want me to buy the challenge. It had taken their customer support ages to reply. They probably had enough buyers at that discount.


I don't use hard stops. My EA uses market orders. Yes, I've deeply learned the rules of at least 10 firms and came to conclusion that they will probably always find a reason to disqualify. I don't fully trust them. I think their only source of income is challenge fees and profitable traders are simply an unwanted byproduct. Because in order to offset the payments to them, the firms need to start copying them on live accounts at some point. But that's a risky business. Why would they engage in it if they have such a nice flow of risk free money as the challenge fees. But I don't know. Maybe I still should try at least a small account at FTMO. What do you think?

Try a demo first at FTMO. Try to be more patient. 🙂

I only use swing accounts, as there is no restriction for overnight holding of positions or news trading. IMO the 1:30 leverage they offer there is sufficient.

 

[AriaS]

Try a demo first at FTMO. Try to be more patient. 🙂

I only use swing accounts, as there is no restriction for overnight holding of positions or news trading. IMO the 1:30 leverage they offer there is sufficient.

But demo is only for testing yourself, isn't it? It doesn't really test the company. They don't care what you do on their demo. What do you think about what I said in the previous reply? Do you think we can trust them?

 

[IlIlIlIlI]

But demo is only for testing yourself, isn't it? It doesn't really test the company. They don't care what you do on their demo. What do you think about what I said in the previous reply? Do you think we can trust them?

The swing account demo is identical to a real swing account. Prices are identical to a real account, but with indices they are different to Darwinex. Very interesting, that pricing .. 😉 It could be a problem if simple trade copiers are used.
The only rules you could violate with a swing are the drawdown rules.

Yes, I trust them. I think they play the most fair game in that market I also know that they might halt markets in very seldom cases for some seconds if the moves are too extrem. These details can only be known by experience.

For the standard account they are cheating as all prop firm that they don't show you if you violated news trading, overnight positions etc. rules.
They have documented that, but that does not help if you don't see it checking new calendar etc.
I think it takes too much effort for them to supervise all demo accounts on theses rules.

 

[AriaS]

The swing account demo is identical to a real swing account. Prices are identical to a real account, but with indices they are different to Darwinex. Very interesting, that pricing .. 😉 It could be a problem if simple trade copiers are used.
The only rules you could violate with a swing are the drawdown rules.

Yes, I trust them. I think they play the most fair game in that market I also know that they might halt markets in very seldom cases for some seconds if the moves are too extrem. These details can only be known by experience.

For the standard account they are cheating as all prop firm that they don't show you if you violated news trading, overnight positions etc. rules.
They have documented that, but that does not help if you don't see it checking new calendar etc.
I think it takes too much effort for them to supervise all demo accounts on theses rules.

I know that the demo is identical, but since it's not the challenge account they don't care what you do on it. On the other hand, when you are on the "real" challenge account they keep an eye on it, and as I understand, you don't even have to break the hard rules for them them to black list you. For example, you might do something that they consider too risky. I heard that even scaling trades can be a problem. FTMO's terms an conditions are quite vague.

p.s. I quoted the "real" because none of their accounts are real, but simulated.

 

[IlIlIlIlI]

I know that the demo is identical, but since it's not the challenge account they don't care what you do on it. On the other hand, when you are on the "real" challenge account they keep an eye on it, and as I understand, you don't even have to break the hard rules for them them to black list you. For example, you might do something that they consider too risky. I heard that even scaling trades can be a problem. FTMO's terms an conditions are quite vague.

p.s. I quoted the "real" because none of their accounts are real, but simulated.

The real disadvantage is that they close it after 14 days and that the profit target is cut by 50%.

They give you the relevant results at the end, so it is not true that they don't care.

To test whether you would pass all objectives is only possible if you add the demo results on a paper if you don't wnat to trade hazardously.

 

[AriaS]

My new back testing algorithm

My strategy has several custom made indicators. Each of them has several parameters, so when you add the basic trading settings like stops, trading direction etc, the number of variants in optimization can reach millions.

Basic info: Swing, 27 pairs, MT5

1. Once per month, or more often, I optimize each pair, 3 months back, by recovery factor. The optimization offers 2.5 million variants, but in fact I stop it at 1-2k when I see around 10-20 variants above RF = 2, that have more than X trades, for some statistical representativeness.

2. My strategy is aiming at universal principles, so that it works better on any curve, but random curve fits still happen. This means they succeeded not due to universality, but due to randomness. One of the reasons this happens: lack of parameter dependency support in MT5 back tester - it tests every possible combination. To filter out the random variants of the chosen 10-20, I perform 2 steps. I will reveal only one of them:

a. I run the 10-20 variants on a different curve. Usually, it's the 3 months prior to the optimization period, but also it might be some other periods of the past. Eventually, for the next month of my live trading I choose the variant with the highest recovery factor, that has more than X trades.