For those who are not programmers but have ideas for creating EAs, "dedicated to idea traders and programmers" - page 7

 
Vita писал (а):

Suppose there is a system that produces trades with a 70% to 30% probability of profit/loss. It does not even matter if we know about it. Suppose there were a hundred profitable trades in a row before the next decision was made. I tend to believe that the probability of the outcome of a future trade does not change, but remains the same - 70/30. Probability of outcome is an intrinsic property of MTS, and as long as the MTS remains unchanged, the probability of the outcome of the trade remains the same (as long as the coin is correct, it will fall 50/50 as long as it is not bent, undercut, etc.). In order to adjust lots "on the fly", I have to believe that MTS suddenly appears or disappears "internal reserves" to increase (decrease) a favorable outcome, just on the basis of previous performance. Or suddenly, after a series of successive losses the market will condescend to me and throw a couple or three profits on my shoulders? The same applies to MTS showing potential profit twice as many losses with equal probability. Previous trade results do not take anything away or add anything to MTS and therefore it will with equal success generate profits twice as many losses.

In my opinion, statistics are enough to convince us that there is no obvious statistical advantage in the market. I suppose it is possible to trade based on the properties of the market.

I completely agree with the fact that the outcome of a future trade is independent of the result of the previous one, provided that the new trade has not yet been opened and the previous one has already been closed.
I do not agree with the marked text. If you can, explain the logic of the conclusion that if you do not change the MTS, the probability remains unchanged.
I believe this is true in the only case - when the market will be shaped by our MTS settings. In all other cases I see no basis for such an assertion.
The worst thing is that the market does not care about anything except supply and demand and the worst thing is that we cannot measure this factor in real terms, But making a forecast we can actually foresee that so much money will come in and so much will go out that it will change our price - it might be easier to hire a shaman because he will be able to hit us on his tambourine. :)

 

The logic behind the idea that "as long as the MTS remains the same, the probability of a deal remains the same" is very simple. The settings have nothing to do with it. Let's take for example a simple MTS that randomly makes a 50/50 buy or sell with equal stops on profit and loss. Whatever the result of this MTS in the past, the profit-loss probability for the upcoming trade of this MTS is still 50/50. No amount of money management will help this MTS become profitable. And until we change the internal logic of buying and selling decisions, the probability of outcomes will not change.

The same is true for the MTS with sophisticated logic - the future profit probability of a deal depends on properties of the MTS itself, and if nothing in it is adjusted "on the fly", then there is no reason to change the lot size. On the other hand, one can easily understand trader's psychology: he or she wants to decrease the lot size in case of incoming losses and increase it in case of incoming profits. But we must distinguish between the desirable and the real. Personally, I find in such behavior a confirmation of the folk wisdom: "That is the wisdom that is easily absorbed by traders who want to be overly prudent.


 
Vita писал (а):

The logic behind the idea that "as long as the MTS remains the same, the probability of a deal remains the same" is very simple. The settings have nothing to do with it. Let's take for example a simple MTS that randomly makes a 50/50 buy or sell with equal stops on profit and loss. Whatever the result of this MTS in the past, the profit-loss probability for the upcoming trade of this MTS is still 50/50. No amount of money management will help this MTS become profitable. And until we change the internal logic of buying and selling decisions, the probability of outcomes will not change.

The same is true for the MTS with sophisticated logic - the future profit probability of a deal depends on properties of the MTS itself, and if nothing in it is adjusted "on the fly", then there is no reason to change the lot size. On the other hand, one can easily understand trader's psychology: he or she wants to decrease the lot size in case of incoming losses and increase it in case of incoming profits. But we must distinguish between the desirable and the real. Personally, I find in such behavior a confirmation of the folk wisdom: "That is the wisdom that is easily absorbed by traders who want to be overly prudent.


Are you saying that the outcome of this trade with a random (50/50) entry is 50/50?
That is, if you do not randomly enter, then the outcome will not be 50/50?

And what do you do with the spread?
Look, we open in different directions with the same number of points on stops. Suppose we open at 1.2800 on the sell, stop at 1.2850 and profit at 1.2750. Where is the equal probability outcome here?
53 points to profit and 47 to loss, what should we write off the 2 spreads on?
 
1

Account: 1 Name: 1 Currency: USD 2006 August 29, 20:08
Closed Transactions:
Ticket Open Time Type Lots Item Price S / L T / P Close Time Price Commission Taxes Swap Profit
7155016 2006.08.28 09:40 balance Deposit 100 000.00
7167328 2006.08.28 20:21 buy 0.10 eurusd 1.2794 1.2776 1.2812 2006.08.28 23:04 1.2776 0.00 0.00 0.00 -18.00
7167341 2006.08.28 20:23 buy 0.10 gbpusd 1.8965 1.8941 1.8989 2006.08.28 21:00 1.8941 0.00 0.00 0.00 -16.80
7167355 2006.08.28 20:24 sell 0.10 usdjpy 117.14 117.32 116.96 2006.08.29 02:47 116.96 0.00 0.00 -1.41 15.39
7167439 2006.08.28 20:32 sell 0.10 eurjpy 149.88 150.12 149.64 2006.08.29 09:52 149.64 0.00 0.00 -0.96 20.57
7167531 2006.08.28 20:40 sell 0.10 usdchf 1.2350 1.2380 1.2320 2006.08.29 06:29 1.2320 0.00 0.00 -1.12 24.35
7167582 2006.08.28 20:45 sell 0.10 eurchf 1.5798 1.5822 1.5774 2006.08.29 12:04 1.5774 0.00 0.00 -0.60 19.48
7167587 2006.08.28 20:45 sell 0.10 audusd 0.7592 0.7616 0.7568 2006.08.29 03:51 0.7616 0.00 0.00 -0.34 -48.00
7167612 2006.08.28 20:47 sell 0.10 usdcad 1.1097 1.1127 1.1067 2006.08.28 22:59 1.1127 0.00 0.00 0.00 -26.96
7167669 2006.08.28 20:54 buy 0.10 gbpjpy 222.14 221.66 222.62 2006.08.29 04:01 221.66 0.00 0.00 1.52 -28.76
7167707 2006.08.28 20:57 sell 0.10 gbpchf 2.3416 2.3464 2.3368 2006.08.29 08:50 2.3368 0.00 0.00 -1.35 27.29
7167871 2006.08.28 21:05 buy 0.10 gbpusd 1.8953 1.8929 1.8977 2006.08.28 22:30 1.8929 0.00 0.00 0.00 -16.80
7167988 2006.08.28 21:14 buy 0.10 chfjpy 94.85 94.55 95.15 2006.08.29 16:25 94.55 0.00 0.00 0.18 -25.65
7168100 2006.08.28 21:19 sell 0.10 nzdusd 0.6381 0.6405 0.6357 2006.08.29 05:15 0.6405 0.00 0.00 -0.78 -48.00
7168143 2006.08.28 21:22 buy 0.10 usdsek 7.2445 7.2145 7.2745 2006.08.29 09:01 7.2145 0.00 0.00 0.72 -41.58
7168171 2006.08.28 21:24 sell 0.10 usdnok 6.2893 6.3193 6.2593 2006.08.29 16:49 6.3193 0.00 0.00 -0.76 -47.48
7168242 2006.08.28 21:30 buy 0.10 usddkk 5.8352 5.8172 5.8532 2006.08.29 05:16 5.8172 0.00 0.00 0.51 -30.95
7168445 2006.08.28 21:52 buy 0.10 usdzar 7.1788 7.1188 7.2388 2006.08.29 08:18 7.1188 0.00 0.00 -0.80 -84.33
7168862 2006.08.28 22:31 buy 0.10 gbpusd 1.8934 1.8910 1.8958 2006.08.29 02:19 1.8958 0.00 0.00 -0.27 16.80
7169244 2006.08.28 22:59 sell 0.10 usdcad 1.1121 1.1151 1.1091 2006.08.29 09:02 1.1091 0.00 0.00 -0.36 27.05
7169358 2006.08.28 23:04 sell 0.10 eurusd 1.2776 1.2794 1.2758 2006.08.29 00:38 1.2794 0.00 0.00 0.77 -18.00
7170212 2006.08.29 00:42 sell 0.10 eurusd 1.2790 1.2808 1.2772 2006.08.29 02:39 1.2808 0.00 0.00 0.00 -18.00
7190407 2006.08.29 19:12 sell 0.10 eurusd 1.2765 1.2783 1.2747 2006.08.29 20:01 1.2783 0.00 0.00 0.00 -18.00
7190433 2006.08.29 19:13 sell 0.10 gbpusd 1.8913 1.8937 1.8889 2006.08.29 20:02 1.8937 0.00 0.00 0.00 -16.80
7190693 2006.08.29 19:26 buy 0.10 nzdusd 0.6413 0.6389 0.6437 2006.08.29 20:00 0.6389 0.00 0.00 0.00 -48.00
7191382 2006.08.29 20:01 sell 0.10 nzdusd 0.6399 0.6423 0.6375 2006.08.29 20:07 0.6423 0.00 0.00 0.00 -48.00
0.00 0.00 -5.05 -449.18
Closed P/L: -454.23
Open Trades:
Ticket Open Time Type Lots Item Price S / L T / P Price Commission Taxes Swap Profit
7167917 2006.08.28 21:08 sell 0.10 euraud 1.6833 1.6923 1.6743 1.6808 0.00 0.00 0.96 19.05
7167878 2006.08.28 21:06 buy 0.10 eurcad 1.4210 1.4150 1.4270 1.4193 0.00 0.00 -0.67 -15.32
7167652 2006.08.28 20:51 sell 0.10 eurgbp 0.6745 0.6769 0.6721 0.6752 0.00 0.00 0.61 -13.28
7168224 2006.08.28 21:29 sell 0.10 usdsgd 1.5773 1.5833 1.5713 1.5756 0.00 0.00 -0.67 10.79
7190476 2006.08.29 19:15 sell 0.10 usdjpy 116.89 117.07 116.71 116.81 0.00 0.00 0.00 6.85
7190493 2006.08.29 19:16 buy 0.10 eurjpy 149.29 149.05 149.53 149.41 0.00 0.00 0.00 10.28
7190497 2006.08.29 19:16 sell 0.10 usdchf 1.2349 1.2379 1.2319 1.2329 0.00 0.00 0.00 16.22
7190522 2006.08.29 19:17 buy 0.10 eurchf 1.5776 1.5752 1.5800 1.5768 0.00 0.00 0.00 -6.49
7190537 2006.08.29 19:18 sell 0.10 audusd 0.7604 0.7628 0.7580 0.7618 0.00 0.00 0.00 -28.00
7190545 2006.08.29 19:19 buy 0.10 usdcad 1.1104 1.1074 1.1134 1.1096 0.00 0.00 0.00 -7.21
7190595 2006.08.29 19:22 buy 0.10 gbpjpy 221.25 220.77 221.73 221.36 0.00 0.00 0.00 6.59
7190616 2006.08.29 19:23 sell 0.10 gbpchf 2.3363 2.3411 2.3315 2.3369 0.00 0.00 0.00 -3.41
7190640 2006.08.29 19:24 buy 0.10 chfjpy 94.70 94.40 95.00 94.74 0.00 0.00 0.00 3.42
7190699 2006.08.29 19:27 buy 0.10 usdsek 7.2328 7.2028 7.2628 7.2154 0.00 0.00 0.00 -24.12
7190708 2006.08.29 19:27 buy 0.10 usdnok 6.3239 6.2939 6.3539 6.3011 0.00 0.00 0.00 -36.18
7190718 2006.08.29 19:28 sell 0.10 usddkk 5.8402 5.8582 5.8222 5.8324 0.00 0.00 0.00 13.37
7190731 2006.08.29 19:29 buy 0.10 usdzar 7.1450 7.0850 7.2050 7.1450 0.00 0.00 0.00 0.00
7191502 2006.08.29 20:01 sell 0.10 eurusd 1.2781 1.2799 1.2763 1.2794 0.00 0.00 0.00 -13.00
7191566 2006.08.29 20:03 buy 0.10 gbpusd 1.8935 1.8911 1.8959 1.8956 0.00 0.00 0.00 14.70
0.00 0.00 0.23 -45.74
Floating P/L: -45.51
Working Orders:
Ticket Open Time Type Lots Item Price S / L T / P Market Price
No transactions
Summary:
Deposit/Withdrawal: 100 000.00 Credit Facility: 0.00
Closed Trade P/L: -454.23 Floating P/L: -45.51 Margin: 2 197.39
Balance: 99 545.77 Equity: 99 500.26 Free Margin: 97 303.01
Details:
Gross Profit: 144.86 Gross Loss: 599.09 Total Net Profit: -454.23
Profit Factor: 0.24 Expected Payoff: -18.17
Absolute Drawdown: 454.23 Maximal Drawdown (%): 454.23 (0.45%)
Total Trades: 25 Short Positions (won %): 15 (40.00%) Long Positions (won %): 10 (10.00%)
Profit Trades (% of total): 7 (28.00%) Loss trades (% of total): 18 (72.00%)
Largest profit trade: 26.69 loss trade: -85.13
Average profit trade: 20.69 loss trade: -33.28
Maximum ($): 3 (65.18) consecutive losses ($): 6 (-204.51)
Maximal consecutive profit (count): 65.18 (3) consecutive loss (count): -204.51 (6)
Average consecutive wins: 1 consecutive losses: 3
Random entry, stop/profit sizes equal, result 1 to 3 so far, amounts unimportant now, only + or -.
 
Daemon:
Are you saying that the outcome of this trade with a random (50/50) entry is 50/50%?
I.e. if you do not randomly enter, then the outcome will not be 50/50?

And what do you do with the spread?
Look, we open in different directions with the same number of points on stops. Suppose we open at 1.2800 on the sell, stop at 1.2850 and profit at 1.2750. Where is the equal probability outcome here?
53 points to profit and 47 to loss, 2 spreads on what to write off?


I don't want to be mesmerised by the randomness of 25 trades. Do you really think your example can be used to indicate a 1 to 3 ratio? You seem to want to believe in that ratio. I, on the other hand, prefer simple and illustrative examples that dissect the point rather than wandering you through three pines. Let me show you a couple.

Oh, and as for the spread, it's not going anywhere. Random MTS is leaking on it. Look closely at the test, which is more or less statistically significant compared to yours. The first report for MTS with stops equidistant by points (if we open with bid, it means that stops are 50 points apart from ask and vice versa). As you see, we have about the same number of wins and losses. The market has the same steps - both to profit and to loss. Here we have an equal probability outcome. Of course, for EURUSD, the average profit trade is lower by about three points than 50, whereas the average loss trade is larger by 3 points.

Random


Symbol EURUSD (Euro vs US Dollar)
Period 1 Hour (H1) 2002.07.24 08:00 - 2006.08.25 00:00 (2001.01.01 - 2006.08.25)
Model All ticks (based on all smallest available periods with fractal interpolation of each tick)
Parameters
Bars in history 25530 Modelled ticks 3895915 Modeling quality 45.60%
Initial deposit 10000.00
Net profit -8980.32 Total profit 66067.00 Total loss -75047.32
Profitability 0.88 Expected payoff -3.18
Absolute drawdown 9041.44 Maximum drawdown 9041.44 (90.41%) Relative drawdown 90.41% (9041.44)
Total trades 2824 Short positions (% win) 1394 (49.07%) Long positions (% win) 1430 (50.70%)
Profitable trades (% of all) 1409 (49.89%) Loss trades (% of all) 1415 (50.11%)
Largest profitable deal 50.60 losing transaction -57.80
Average profitable deal 46.89 losing trade -53.04
Maximum number continuous wins (profit) 11 (517.00) Continuous losses (loss) 9 (-476.04)
Maximum Continuous Profit (number of wins) 517.00 (11) Continuous loss (number of losses) -476.04 (9)
Average continuous winnings 2 continuous loss 2



And here is the report with stops of equal profit-loss values. Note that average profit-loss has become equal, but the ratio of profitable-loss trades has changed. The market now goes farther to profit on the spread, and closer to loss on the spread. Therefore, losses have become more frequent, so this MTS will lose the spread in the same way as the previous one.

Regardless of previous trades, lot sizes and what you believe, each of these simple MTS will generate profit/loss in the future with the same probability. But by playing with the size of stops and lots, mixing them on different pairs, it is very easy to fool yourself into believing that such shamanic manipulation will change the probability of future profits/losses.

Strategy Tester Report
Random


Symbol EURUSD (Euro vs US Dollar)
Period 1 Hour (H1) 2002.07.24 08:00 - 2006.08.25 00:00 (2001.01.01 - 2006.08.25)
Model All ticks (based on all smallest available periods with fractal interpolation of each tick)
Parameters
Bars in history 25530 Modelled ticks 3895915 Modeling quality 45.60%
Initial deposit 10000.00
Net profit -9337.08 Total profit 62639.80 Total loss -71976.88
Profitability 0.87 Expected payoff -3.46
Absolute drawdown 9406.24 Maximum drawdown 9406.24 (94.06%) Relative drawdown 94.06% (9406.24)
Total trades 2695 Short positions (% win) 1329 (45.75%) Long positions (% win) 1366 (47.44%)
Profitable trades (% of all) 1256 (46.60%) Loss trades (% of all) 1439 (53.40%)
Largest profitable trade 54.32 losing transaction -53.84
Average profitable deal 49.87 losing trade -50.02
Maximum number continuous wins (profit) 11 (548.32) Continuous losses (loss) 11 (-551.92)
Maximum Continuous Profit (number of wins) 548.32 (11) Continuous loss (number of losses) -551.92 (11)
Average continuous winnings 2 continuous loss 2


For example, I use MTS with a random input to test the trading code (not the logic of decision making). If I obtain the figures shown in my reports, it means there is a good chance that the trading code does not contain gross errors. I must give credit to the MT tester that it manages testing of simple MTS perfectly. I wish you do the same.

 

What makes you think the testing stopped after 25 trades?

What I had on hand is what I posted. I'm sorry I don't have a recipe. I'll wait for yours.

 
Daemon:

What makes you think the testing stopped after 25 trades?

What I had on hand is what I posted. I'm sorry I don't have a recipe. I'll wait for yours.


It's very kind of you to post what you had on hand. I, on the other hand, commented on what you posted, what you wrote and defended my point of view which you disagreed with. I don't read your mind. Note that I assumed nothing about the fate of your testing. However, I am inclined to believe that you have nothing serious to back up your point.
 

What are you defending? that the result you promised should be 50/50? Where is it? I don't see proof of it, although I put it for the test myself, it has leveled the ratio of 40 profitable 60 losing or 2/3, but it is not 50/50, don't you get it? I'm sitting now with a mathematician discussing this case to prove YOUR point. The conclusion is that for 50/50 deal with random entry it is necessary to have a spread in the 4th digit category, ie simply put, in order to have probability close to 0.50 with equal stops and profits it is necessary for the side to have a size of 1000 points, you mean? What MM may be discussed at such positions and how long you have to wait for the result? Please specify.
As far as I've seen no proof that random input can give 50/50 and non-random input can give 50/50, all we have is talk.
With random entry it is clear, it is random. What about non-random entry? :)
It would be interesting to discuss this topic with a person who can prove something with his own example of success, while I will not listen to proofs and axioms of the like and thus believe unconditionally. If you do not have any real success, be kind and prove at least on paper, we will check together.

 
Daemon:

What are you defending? that the result you promised should be 50/50? Where is it? I don't see proof of it, although I put it for the test myself, it has leveled the ratio of 40 profitable 60 losing or 2/3, but it is not 50/50, don't you get it? I'm sitting now with a mathematician discussing this case to prove YOUR point. The conclusion is that for 50/50 deal with random entry it is necessary to have a spread in the 4th digit category, ie simply put, in order to have probability close to 0.50 with equal stops and profits it is necessary for the side to have a size of 1000 points, you mean? What MM may be discussed at such positions and how long you have to wait for the result? Please specify.
As far as I've seen no proof that random input can give 50/50 and no proof that non-random input can give 50/50, all we have is talk.
With random entry it is clear, it is random. What about non-random entry? :)
It would be interesting to discuss this topic with a person who can prove something with his own example of success, while I will not listen to proofs and axioms of the like and thus believe unconditionally. If you do not have any real success, be kind and prove at least on paper, we will check together.


To prove my point I have posted the results of an EA randomly opening to buy or sell. The spread is easily accounted for and the 50/50 ratio does not change in any way. My example clearly shows on which period, currency and time frame this result is obtained. I don't need any warrantless faith. Anyone can check it. Your statement about 40/60 will remain a myth that cannot be verified for all those present until you indicate where and under what conditions you have achieved such a result. Again, can you offer us anything that can confirm the 40/60 hypothesis? At least test parameters so that the 40/60 hypothesis can be tested regardless of your beliefs?

Moreover, I can tell that in other intervals with other currencies and timeframes with number of deals around a thousand we obtain the same 50/50 results as I have stated. 50/50 is a property of the market. At any time from a fixed position the market reaches equidistant points with equal probability - 50/50. There is no need to remove stops by 1000 pips. To verify the 50/50 hypothesis, a 1 point deletion is enough, but the number of checks should be increased to at least 1000. Repeat the same for 2 pips. For 3 pips and so on. The result will invariably indicate 50/50. As a salute to your mathematician - I pass on the word "statistics".

Non-random entry is out of the question. The market does not give a statistical advantage for the MTS I mentioned.
 

The difference is that you are playing and I am working. The results of testing in the STRATEGY TESTER of similar EAs, I can also lay out, but because I respect the local people and you among others, I'm testing in real time. Somewhere here on the forum I think I told you how I tested an EA on a new build. And it opened a position, worked on a stop and showed a profit, stop loss was not modified, what kind of proof can we talk about in this tester seriously? At first I took your arguments with example from testers as a joke, but now I see that all this is not so funny. Do you want someone to analyse your EA's trades or do you want to do it yourself?

Simply put, the quality of your simulation is 45% as shown in your tester, while my simulation is 100% :) Is that an argument?