Another bit of Martingale lunacy | Page 2 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Martingale, Anti-martingale, and Compounding 40 replies

Martingale, Reverse Martingale, Modified Martingale, Maths 1 reply

Martingale Calculation EA (also reverse martingale) 3 replies

How to change this Martingale to Reverse Martingale ? 3 replies

Martingale vs. Non Martingale (Simplified RoR vs Profit) 1 reply

Attachments: Another bit of Martingale lunacy
Exit Attachments

Another bit of Martingale lunacy

  • #21
  • Jun 9, 2011 7:36am Jun 9, 2011 7:36am
I had an idea for a modification for this EA, I don't know how easy it would be to implement but wanted to share it to see what people thought.

It is trying to capitalize on psychological levels - primarily the 00 and 50 round number levels. When price hits these levels it is usually rejected fairly strongly OR breaks through fairly strongly. In other words there is movement around these levels (which is required for this EA to perform positively).

Could a SELL order be placed as price approaches these 00 & 50 levels (say gets to 1-2 pips away) from below the level. The price will either be rejected from the level for a good 10 pips, or it will break-through (usually strongly) conversely gaining the 10 pips it just lost.

The opposite would work for a BUY order as price approaches these levels from above.

Like I said, I'm not sure how easy this would be to implement but it seems like a good idea on paper.

Many thanks

Jimmy
The best trade I ever made was my SIN for Jesus' RIGHTEOUSNESS.
  • #22
  • Jun 9, 2011 8:31am Jun 9, 2011 8:31am
Hmmmm. Early days, but I am coming to like this bot. I am even considering changing its name and dropping the 'lunacy' bit.

Latest update in post 1:

 

  1. added semi-automated initial trade direction calculation.
  2. added the option to stop trading for the session once a tp has been hit.

Details in post 1.

  • #23
  • Jun 9, 2011 8:33am Jun 9, 2011 8:33am
Thanks Steve for the new updates. I was doing this manually.
Looks like I don't need to miss my sailing anymore--

Great work again,



Quoting SteveHopwood
Hmmmm. Early days, but I am coming to like this bot. I am even considering changing its name and dropping the 'lunacy' bit.

Latest update in post 1:

 

  1. added semi-automated initial trade direction calculation.
  2. added the option to stop trading for the session once a tp has been hit.

Details in post 1.

  • #24
  • Jun 9, 2011 8:34am Jun 9, 2011 8:34am
Quoting SteveHopwood
Hmmmm. Early days, but I am coming to like this bot. I am even considering changing its name and dropping the 'lunacy' bit.

Latest update in post 1:

 

  1. added semi-automated initial trade direction calculation.
  2. added the option to stop trading for the session once a tp has been hit.

Details in post 1.

Yes, I've even removed my earplugs
  • #25
  • Jun 9, 2011 8:41am Jun 9, 2011 8:41am
I am preferring using 15 pips TP and SL in volatile markets.
You may all consider this slight modification.

Best,
  • #26
  • Jun 9, 2011 8:42am Jun 9, 2011 8:42am
Quoting mysticgenie
Interesting developments so far.
Like Steve's idea of BB filter.
Basically as long as market moves more than 10 pips upto 20 pips you will get your profit ( if euro is trending up). If euro is down, you will lose first position, but the next sell position will bring you the profit as long as it moves 20 pips.
You can just manually turn on or off this EA during market over lap hours.
Turn on during late Asia early lodon open, and late london and early US market overlap. these times market move significantly no matter what as long as they are open....
Hey Mystic,

Nice 2 see u active again.

just out of curiosity were u ever able to repeat ur 1k to 100k feat?
  • #27
  • Jun 9, 2011 9:12am Jun 9, 2011 9:12am
The problem with any Martingale strategy is that the number of wrong trades before your account is blown is surprisingly small.

Even if only risking 1% of your account on the first trade by trade 6 you run out of runway (cumulative losses of 1%, 3%, 7%, 15%, 31%, 63%), meaning trade 7 has to have a stake of 64%, which is a problem because you only have 37% of the original pot still intact.

And even 0.1% only buys you another 3 trades

For the vast majority of us mere mortals, risking 0.1% of our account isn't going to make us very rich!

The chances of a successful trade are higher if the target is low, but the lower the target the easier a loss also becomes.

As an example, I spreadbet with GBPUSD typically with a 2 pip spread. Now if I'm shooting for 100 pips then the 2 pip spread becomes fairly insignificant. But if I'm shooting for 10 then I have to make 12 pips to close the trade profitably but can only afford an 8 pip move before my 10 pip SL kicks in. So the market has to move 50% further to hit my TP than it does to hit my SL.

I've tried demo variations of Martingale on GBPUSD using 10 pips, 20, 30, 40, all the way to 100, and every single one resulted in a blown account at some point between Jan 2009 and June 2011

Andy
  • #28
  • Jun 9, 2011 10:49am Jun 9, 2011 10:49am
Quoting Andy B
The problem with any Martingale strategy is that the number of wrong trades before your account is blown is surprisingly small.

Even if only risking 1% of your account on the first trade by trade 6 you run out of runway (cumulative losses of 1%, 3%, 7%, 15%, 31%, 63%), meaning trade 7 has to have a stake of 64%, which is a problem because you only have 37% of the original pot still intact.

And even 0.1% only buys you another 3 trades

For the vast majority of us mere mortals, risking 0.1% of our account isn't going to make us very rich!

The chances...
All of which is true, but you haven't really read post one and recent developments have you?

People will only blow their account this way if they misunderstand what mystic is getting at - as I did when he first contacted me.

The safeguards against account-killing are all here for those that read them and take notice.

  • #29
  • Jun 9, 2011 10:50am Jun 9, 2011 10:50am
Quoting mysticgenie
I am preferring using 15 pips TP and SL in volatile markets.
You may all consider this slight modification.

Best,
Cheers. I will give it a whirl tomorrow.

  • #30
  • Jun 9, 2011 11:01am Jun 9, 2011 11:01am
Quoting SteveHopwood
All of which is true, but you haven't really read post one and recent developments have you?

People will only blow their account this way if they misunderstand what mystic is getting at - as I did when he first contacted me.

The safeguards against account-killing are all here for those that read them and take notice.

I have read them, but the fact remains that a run of 6 trades will lose 6% of your account if you just take the loss but will blow your account if you use a Martingale strategy

Having an edge for the first trade doesn't change this fundamental truth

Or maybe I am missing something?
  • #31
  • Jun 9, 2011 5:10pm Jun 9, 2011 5:10pm
There is a bug in your code, by fixing it the result could be even better (or worse ). Your int TrendDirection() function returns 0 if neither BUY or SELL conditions are met. The problem is OP_BUY also happenes to have value of zero, so when niether conditions are met you basically say "BUY".

Quoting MaryJane
Pure, unlimited marty is a killer. But if there's an underlying strategy that's profitable in itself and has a reasonable profit/loss distribution pattern, a limited marty can actually (surprisingly) improve the expectancy. Let's forget the buy/sell/buy or buy/buy/buy coin-toss for a while (there's no edge). Take some KISS "strategy" like this:
  • #32
  • Jun 10, 2011 1:32am Jun 10, 2011 1:32am
Quoting Orest
The problem is OP_BUY also happenes to have value of zero...
Ooops didn't realize that. Still learning, thanks! I ran a quick BT (not on tick data) and yes, results are slightly worse if fixed. But the rsi filter is a very crude method, you get pretty much the same result with just rsi<50/>50. The EA mod I posted wasn't mentioned to be traded with, i only wanted to point out how important it is to have some statistical edge with the entries and in this case it was tha fact that eurusd tends to make its daily highs and lows around gmt+1. mikkom published this in 2009 and nothing changed since then. The Peter Crowns' direction principle as introduced by Steve in post#1 is of the same kind.

Quoting Andy B
I have read them, but the fact remains that a run of 6 trades will lose 6% of your account if you just take the loss but will blow your account if you use a Martingale strategy

Having an edge for the first trade doesn't change this fundamental truth

Or maybe I am missing something?
Just a word to all the anti-martingalists: this is (again) not a set-and-forget EA! mystic's strategy is a volatility exploit and the ea has to be used manually, as a tool. And from what i have seen, mystic is a very good and experienced trader but also he uses a lot of discretion; he's able to feel the market and probably well versed in managing risk
MarginTrader All Time Return: na
  • #33
  • Jun 10, 2011 4:44am Jun 10, 2011 4:44am
Quoting MaryJane
Pure, unlimited marty is a killer. But if there's an underlying strategy that's profitable in itself and has a reasonable profit/loss distribution pattern, a limited marty can actually (surprisingly) improve the expectancy.
Increasing stake size when probabilities are increasingly in your favor will improve return (ask any blackjack pro). However, increasing stake size merely to recover recent losses has no statistical basis; market behavior and probabilities are obviously not going to change based on one insignificant retail trader's recent performance.

One major problem I have with doubling is that the stakes quickly escalate to a level that can't be justified by probabilities on offer. After 4 losses, for example, one is now trading 16x the original stake, but the expectancy is very unlikely to be around 16x as great as it was originally. Now the trader needs 15 consecutive wins at 1 unit to atone for the 4 losses. And it also raises the question as to what is a safe level of staking for a given account size, in the context of avoiding ruin.

It can be shown mathematically that staking systems in themselves can not improve expectancy (some time ago I posted an XLS that allowed testing of various staking systems). Even in a +EV situation, simply using a suitable flat bet size will deliver the same overall expectancy as any other staking system.
  • #34
  • Jun 10, 2011 5:47am Jun 10, 2011 5:47am
Steve - have you thought of linking this (and giving it an edge) to Pippo or Ibuks methodolgies. Now that might be be quite interesting.
  • #35
  • Jun 10, 2011 5:57am Jun 10, 2011 5:57am
Quoting Toccata
Steve - have you thought of linking this (and giving it an edge) to Pippo or Ibuks methodolgies. Now that might be be quite interesting.
No I hadn't, but it is a thought. Counter-trend the initial trade at the extremes; go with the daily open direction thingy when in the middle.

I am not sure it is necessary when we are only looking for a 10 pip movement, but am not dismissing the idea.

Any further thoughts on this, guys?

  • #36
  • Jun 10, 2011 8:39am Jun 10, 2011 8:39am
Quoting hanover
One major problem I have with doubling is that the stakes quickly escalate to a level that can't be justified by probabilities on offer.
I meant that doubling (or other kind of multiplying, like various 'recovery' methods do) the stake to a very limited extent of steps can be justified only if there are some odds of reasonable profit/loss distribution pattern with the given strategy, based on past performance. And it naturally becomes invalid and contraproductive as soon as the system performance degrades over time as the market changes.

I was actually surprised that allowing 2 steps (1-2-4; then take the loss and reset to 1) increased the gain over 400+ trades by a factor of 2.16 while max dd went up 1.44x and relative dd 1.72x compared to flat bet size. Allowing one more step (1-2-4-8; loss) made the dd figures unreasonable. Maybe the dd calc in the tester is flawed.. and almost certainly my understanding of expectancy is flawed too . Thanks for reply and insight!
MarginTrader All Time Return: na
  • #37
  • Jun 10, 2011 9:04am Jun 10, 2011 9:04am
Quoting MaryJane
I was actually surprised that allowing 2 steps (1-2-4; then take the loss and reset to 1) increased the gain over 400+ trades by a factor of 2.16 while max dd went up 1.44x and relative dd 1.72x compared to flat bet size. Allowing one more step (1-2-4-8; loss) made the dd figures unreasonable. Maybe the dd calc in the tester is flawed.. and almost certainly my understanding of expectancy is flawed too . Thanks for reply and insight!
Manual backtesting of a grid trading strategy that I've used on GBPUSD over 30 months of data showed the same thing: 1-2 works better than 1, and 1-2-4 works even better, but 1-2-4-8 resulted in a significant decline in profits
  • #38
  • Jun 12, 2011 10:40am Jun 12, 2011 10:40am
Quoting Andy B
Manual backtesting of a grid trading strategy that I've used on GBPUSD over 30 months of data showed the same thing: 1-2 works better than 1, and 1-2-4 works even better, but 1-2-4-8 resulted in a significant decline in profits
Just curious. Since you all discovered that martingale works well up to 3rd level. Did you figure out the "grid"? How do you determine the best grid setting?

I mean, how big (or small) your grid also affects the effectiveness of martingale. For example, a 5 pips grid may be too tight for it to handle any big move. On the other hand a 50 pips grid may be too large to catch the retracement. Again, how do you determine the best grid setting? Or is it a trial and error thing?
  • #39
  • Jun 12, 2011 11:09am Jun 12, 2011 11:09am
Quoting delorean
Just curious. Since you all discovered that martingale works well up to 3rd level. Did you figure out the "grid"? How do you determine the best grid setting?

I mean, how big (or small) your grid also affects the effectiveness of martingale. For example, a 5 pips grid may be too tight for it to handle any big move. On the other hand a 50 pips grid may be too large to catch the retracement. Again, how do you determine the best grid setting? Or is it a trial and error thing?
I tested 10, 25 and 50 and found 25 to be a reasonable compromise
  • #40
  • Jun 13, 2011 9:36am Jun 13, 2011 9:36am
Latest update in post 1. Please see 'Pivot'.

0 traders viewing now