Automatic Loss Recovery System (ALR) - read this first

[Kruspe]

This could be solved by dynamically changing TP. With every new opened position will EA calculate BE/ and profit zone and TP will be moved a few pips far from ALR zone to cover spreads and swaps (last opened order will cover it). And this distance will be bigger every day because of swaps. And it seems to be dangerous. What do you think ?

K.

[Dewey McG]

Important update on ALR calculator

This is important for calculating risks and expectancy of profit. I realized a mistake I was making before. While it is very reasonable to experiment with different win rates for the initial trade, once price goes against our position and we begin ALR we have to look at it differently. If we are using a 150 pip tp and 50 pips to another ALR trade, then we are really looking at a 25% win rate on all ALR trades after the initial trade. So even if your strategy would normally give you 70% winners, once to moves against you and the ALR cycle begins everything changes. Now one could argue if your system is trend trading VSA or price action based etc. that you will still be right more than wrong I figured we should be as conservative as possible and use the 25% number. As you use more turns, odds against hitting a stop loss still remain small, but we need to use these numbers. New ALR calculator is attached and on post 1.

[Dewey McG]

OK, I ran some numbers comparing the 3 sets of numbers we have. First of all, I need to repeat what Steve told me and that is that Gary is a genius with numbers and we should all be immensely grateful for his input:

Gary's 1st set of numbers which I am using in the ALR calculator

vs.

Dustin's numbers from his product which was used by Klotsh

vs.

Gary's new set of numbers where we DECREASE lots sizes after a certain point to reduce over all risk

As you can see by the PICs, Gary1 is much better than Dustin's numbers in almost every aspect.

Now the interesting part. Gary2 has many advantages over Gary1, but also a lot more losses. I think we can come up with something in between that will reduce the losses further. It might take me a few more days to see what I can do as we have out of town company arriving.

[Dewey McG]

I chose the sizes that "just barely" got the ALR trades into breakeven/profit. I believe those numbers will result in the slowest growth, assuming you don't want to accept any losers in the ALR turns. Larger sizes would produce more profit from the losers, but the size will grow much faster. Eventually, if the market keeps going against you, you'll have to bail out and accept a Martingale blowup.

Here's an alternate idea: rather than ever accepting a Martingale blowup, you can set the thing up so it gradually unwinds the position.

In this example, you go long X for the first trade. If that fails, you go short 2*x-.01, so now you're net short X-.01. Then each step you reduce the size a bit, reducing the net exposure until you ease out of the trade.


So let's assume the original trade has a 25% chance of hitting the TP. (That's the chance of a random entry hitting a 3:1 TP.) 25% of the time your initial trade wins, and you win 15.00. 25% of the remaining 75% of the time you reverse once and win 8.50, and so on, gradually reducing the position size until you unwind the whole thing. You NEVER have a Martingale blowup this way.

However if you calculate the probabilities of making N turns (assuming the 25% win rate) and figure how often you make 15.00, 8.50, etc -- then the long-term average profit per "trade and N ALR reversals" is ... zero. And that's before costs.

You just can't spin straw into gold.

Dewey, I believe you've been working with a model that risks a limited blowup on rare occasions, yes? I VERY VERY STRONGLY suspect it does the exact same thing, over a longer timeframe. You make lots of trades showing lots of small profits, and then the limited blowups give all those profits back. You might get lucky and make some profits before you quit, or you might get unlucky and take a bunch of blowup losses right at the start. But over the very long run, I suspect the net profits will be zero -- before costs.

TANSTAAFL...
Gary

I really like this idea but think it needs a little tweaking. I still think your first set of numbers would work very well, but this idea could prove to a home run.

[madpipa]

I just ran a backtest on GBPAUD using an Account = $ 1,000,000 & SL = 15, TP = 45. Started with a paltry 0.01 lot size & asked for a minimum break even trade. It survived the last 12 months, and did have a small $1,232 win. But it did have a couple of hiccups!

At one time it placed a series of 33 trades before it finally got out of trouble. The largest lot size was 90.44 full lots! That is $900+ per pip! Might be fine for some my account won't stretch that far.

Unless we can find some satisfactory way of controlling that large lot size blowout then worrying about little things like making spread or a small profit from each trade is redundant. Don't forget the thread's title:
Automatic Loss Recovery

This isn't supposed to be a trading system in itself - more a way to recovery from taking a loss when your normal trade system suffers a "loss".

Cheers,
Mick

[jacobmostert]

Hi Wealthmaster,

thanks for pointing that out. I should have divided the C parameter by 10 assuming we are using standard lots because the pip value of std lot = 10 $ not 1 $.

I corrected this in the sheet and it should be ok now. Find attached below.


until we agree on the formula, is anyone keen to start writing the EA code? Static values can be used in the table that holds the trade sizes for testing.

Cheers!

Hi kotsh

Please have a look at your spreadsheet in columns B13:B21 and B27:B35.

B12 reads:
((C25+$C$2)*($C$5+$C$4)+0.1*$C$6))/($C$7+$C$4))-C11

and B13 onwards reads:
((C26+$C$2)*($C$5+$C$4)+$C$6))/($C$7+$C$4))-C12

Same for B26 and B27 onwards.

Seems you only made the correction for the first line in each column. If one correct all the lines, the position sizes appears to be correct - and much lower!. Also, your nett positions are not calculated correctly; you should add positions in the SAME directions and subtract those in the OPPOSITE direction.

With the corrections your spreadsheet's results seem to be similar to Gary's.

Regards

[kotsh]

Hi jacobmostert,
thanks for pointing these errors out!

Seems you only made the correction for the first line in each column. If one correct all the lines, the position sizes appears to be correct - and much lower!.

yes, you are right, I forgot to drag down the formula to apply to all cells.

Attached is corrected version.

ALR_kotsh_v2.1.xlsx

Also, your nett positions are not calculated correctly; you should add positions in the SAME directions and subtract those in the OPPOSITE direction.

As for this part, yes, that is what the excel formula does, but I am also taking into consideration the initial trade in the net position calculation:
C12-D25-$C$2

Does that clear it up or is there something else that I am missing?

[jacobmostert]

Hi jacobmostert,
thanks for pointing these errors out!

yes, you are right, I forgot to drag down the formula to apply to all cells.

Attached is corrected version.


As for this part, yes, that is what the excel formula does, but I am also taking into consideration the initial trade in the net position calculation:
C12-D25-$C$2

Does that clear it up or is there something else that I am missing?

Hi kotsh

Regarding the nett positions, I see what you are doing... However, I think there is still a small error in the spreadsheet equation in the column D12:D21. Note that in the corresponding column D26:D35 the equation is: =C12-C25-$C$2 while in D12:D21 it is: =C13-D26-$C$2. I believe it should be C13-C26-$C$2.

A further suggestion would be to fix the position sizes in the spreadsheet to two decimals because I think most brokers out there only allow down to 0.01 lots. You can use something like the FIXED() function to actually truncate the values and prevent Excel from using all the decimals - although you might want to round the number up or down before fixing it.

Regards

Jaco

[kotsh]

Regarding the nett positions, I see what you are doing... However, I think there is still a small error in the spreadsheet equation in the column D12:D21. Note that in the corresponding column D26:D35 the equation is: =C12-C25-$C$2 while in D12:D21 it is: =C13-D26-$C$2. I believe it should be C13-C26-$C$2.

Yes, you are totally right! Again

Now the exposure looks much better!

corrected version attached

ALR_kotsh_v2.2.xlsx

A further suggestion would be to fix the position sizes in the spreadsheet to two decimals because I think most brokers out there only allow down to 0.01 lots. You can use something like the FIXED() function to actually truncate the values and prevent Excel from using all the decimals - although you might want to round the number up or down before fixing it.

Yes, I am aware of that. You can also just select the cells --> right-click format and then chose to display only 2 decimal digits which will round up/down the numbers. I left it on purpose to see what the actual value before rounding is.

We can add an EA input that gives the option to round the position sizes up, down or normal.

The rounded position sizes will have the effect that you need slightly more/less pips to reach net profit of 0 (or predefined value)

In all cases the EA should close the positions based on net profit and not TP levels, so it is not really an issue.

[Dewey McG]

Good work Kotsh. Just had a quick peak and it does look much better!

I won't be back until late, but if you want to have a stab I do want to look into Gary's idea of getting to a certain number of turns (my guess would be somewhere between 6- 8 turns) and then drawing down the lot size. This would allow small losses but reducing the exposure even further. This might sound counter intuitive, but it actually might make this strategy a lot more profitable, still making sure almost all series of trades close in profit or just above break even, while reducing the risk even further.

Food for thought.