Is it better to increase lot size after win or after loss? | Page 2 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Which would you prefer: increasing lot size or TP size? 24 replies

At what lot size/account size do you have scale issues? 39 replies

Lot size, contract size, max position ? 1 reply

BUY run profit 1 lot, SELL LIMIT 2 LOT down again n BUY 1 lot - EA Reqd 5 replies

Help: Need code to increase lot size per $1000 increase in account 4 replies

Attachments: Is it better to increase lot size after win or after loss?
Exit Attachments

Is it better to increase lot size after win or after loss?

  • #21
  • Apr 29, 2012 7:20pm Apr 29, 2012 7:20pm
[quote=lolik2020;4429151]Hi,....is it Better 2 increase lot size after win or after loss?
.....quote]

Just fund your account regurarly then doesn't matter when you increase lot size....
The Market pays you to be disciplined
  • #22
  • Sep 8, 2012 9:30pm Sep 8, 2012 9:30pm
Quoting smjones
6. This was a completely random walk coin toss with a margin of error of less than 1/100,000 of 1 percent pure random.

7. I conclude that the reason that this is profitable even for a random coin toss is because this is taking advantage of very short term trends in random data. It is betting more when there is multiple heads and less when there is not and is stopping each of the sessions at a 8:1 reward to risk ratio.

8. This would suggest that there are periods of non randomness in larger areas of random data and would also suggest that increasing...
Scott, I guess you may never see my post here, as I note that this thread is old and you haven't visited the forum for several months. But I will proceed anyway.

I don't understand your points 6, 7 and 8.

'Trends' will occur in random-walk data as a matter of course, i.e. sequences of heads or tails in a fair coin toss. But (and assuming I understand gaming theory correctly) not in such a way that an MM system could exploit them profitably. The logic being this: in a 'perfect' 50/50 random walk, every trial is independent, hence every HH or TT sequence is equally as probable as a HT or TH sequence. By induction, we can show that this likewise holds true for HHH, HHHH, and so forth, over any number of trials.

The attached XLS (which I've posted many times before) confirms that, over 6 trials, then assuming all possible 2^6=64 combinations are equally likely (which they must necessarily be, if the coin is fair), that doubling the stake following a win, and halving it following a loss, generates a result of zero expectancy, which is exactly the same as for consistent staking, or indeed any other MM strategy that you might wish to test.

For your system to show profit (and I believe we can safely assume that 100 million tosses rules out any kind of 'luck') then the trends must be (on average) longer than those that would be created by a perfectly random generator. Hence I suspect that the generator is somehow flawed.

Extending the implications of your result, this suggests that a game like roulette could be beaten, or at least the house edge of -5.26% reduced, by using a MM strategy. Yet, to the best of my knowledge, apart from defective roulette wheels, nobody has ever achieved this.

What generator did you use in your tests?

Thanks,
David
Attached File(s)
File Type: xls Martingale comparison testing.xls   115 KB | 501 downloads
  • #23
  • Edited 11:09pm Sep 8, 2012 9:54pm | Edited 11:09pm
Quoting hanover
Scott, I guess you may never see my post here, as I note that this thread is old and you haven't visited the forum for several months. But I will proceed anyway.

I don't understand your points 6, 7 and 8.

'Trends' will occur in random-walk data as a matter of course, i.e. sequences of heads or tails in a fair coin toss. But (and assuming I understand gaming theory correctly) not in such a way that an MM system could exploit them profitably. The logic being this: in a 'perfect' 50/50 random walk, every trial is independent, hence every HH or TT...

2 things :

1) why are you reviving an old dead thread? odd.

2) my cursory read of the first post suggests to me that his results are probably spot-on and perfectly correct due to :

a) there is no negative edge in flipping a coin, unlike the zeroes in roulette and spread in retail forex.

b) he is cutting his winners, and cutting his losers. he isn't just running 4 million flips and counting up the final tally. there are going to be times where he wins and loses and grinds away to a losing 'game' where he resets back to the original stake sizing. there are also going to be times where he wins and loses but rockets up towards his max win amount condition and then resets. this suggests to me that there are going to be times where he loses at a small stake, and times where he wins at a huge stake.... but where he resets after a won game that get followed by a new game which starts at a losing flip (and so cut his winner perfectly before it got nailed by a bad flip edit: bad flip streak i mean /edit). similarly with the losing games that get cut and which then proceed to start the next game on a winning streak which exponentially grows and hits the win state. it takes less win streaks to hit a jackpot than losing streaks to hit the loss state... so riding a winning trend cashes in over and over and quickly, but hitting a peak and riding down a losing trend takes longer to lose less. the rest of the time washes-out even and matters none due to no costs to 'play'. this is my understanding of why he mentions it is merely a case of capturing short term trends in random data. pyramiding would fair a lot better in the real world if there was no spread.

without thinking too hard about this it seems he's maximizing the full rides up from trough to peak of winning games and making sure to cash-in regularly and locking-in bank, while minimizing the impact of riding peak to trough streaks. everything else in the middle is neither here nor there.


err, or to put it another way... after every loss is not necessarily another smaller stake in the next flip (ie resetting after a hitting loss limit), and after every win is not necessarily a bigger stake in the next flip (ie resetting after hitting the win limit). therefore, a losing streak doesn't put one progressively into smaller stake sizing that leads to crappy win amounts once the streak ends, just as a winning streak doesn't get busted by some crappy losses at a huge stake size once the win streak ends.

.....

that's my cursory take on it, and is probably totally rubbish. i've probably missed something obvious.

i know of smjones, and know that he has touched on some stuff others have no idea even exists. i'd tend to give him the benefit of the doubt on just about anything he proposes... doesn't mean he didn't make a mistake somewhere, but still.
Congratulations ScalaFX for coming out of the closet!
  • #24
  • Sep 9, 2012 2:44am Sep 9, 2012 2:44am
Quoting nubcake
2 things :

1) why are you reviving an old dead thread? odd.

2) my cursory read of the first post suggests to me that his results are probably spot-on and perfectly correct due to :

a) there is no negative edge in flipping a coin, unlike the zeroes in roulette and spread in retail forex.

b) he is cutting his winners, and cutting his losers. he isn't just running 4 million flips and counting up the final tally. there are going to be times where he wins and loses and grinds away to a losing 'game' where he resets back to the original stake sizing....
Nub,

Thanks for your response. I stumbled on this thread fortuitously, and as 'casino-type math' is a pet topic of mine, couldn't resist adding my 2c. I know this topic's been covered many times before........

I have great respect for Scott; we exchanged many posts and PMs back in the earlier days of FF (c 2007).

I agree 100% with your point about there being no negative edge in a coin toss (unlike FX's spread and roulette's zero). But a 50/50 event should generate zero expectancy, no matter what MM is used. I originally believed (actually, hoped) that it was possible to create an edge using MM, and spent a long time trying to prove it. In the end I gave up (I decided I was being fooled by a kind of Gambler's fallacy), and the XLS that I posted summarizes the reason why. It fascinates me how, no matter what staking strategy is used, i.e. increase/decrease bet size following a win/loss, or a streak of wins/losses, that when that MM strategy is projected over every possible combo of equally probable outcomes, the overall expectancy always neatly balances out to zero.

I'll create a simulator, using Scott's rules, and see what I can come up with. If it's possible to devise a MM that's capable of offsetting the negative edge imposed by spread, then it's surely a worthwhile exercise. But I have a feeling it'll be a waste of time.

David
  • #25
  • Sep 9, 2012 3:08am Sep 9, 2012 3:08am
1) trade small
2) no SL
Scenario: you have 2 open trades:
3) lucky you you have winner
a) keep it, close it, add to it ... it's a winner you are making some $$
b) do it again
4) back luck your other trade is now down 500 pips
a) Use every reversal to reduce position.
b) Lucky you, you waited it out and you bought at a reversal. Up 100 pips ==> close for 100 pips profit and close a 100 pips loss = loss 400 pips.
Start again at 3)
  • #26
  • Sep 9, 2012 3:38am Sep 9, 2012 3:38am
Quoting hanover
Nub,

Thanks for your response. I stumbled on this thread fortuitously, and as 'casino-type math' is a pet topic of mine, couldn't resist adding my 2c. I know this topic's been covered many times before........

I have great respect for Scott; we exchanged many posts and PMs back in the earlier days of FF (c 2007).

I agree 100% with your point about there being no negative edge in a coin toss (unlike FX's spread and roulette's zero). But a 50/50 event should generate zero expectancy, no matter what MM is used. I originally believed (actually,...
you would think a 50/50 would result in even steven... but some dark mathemagics can produce some 'interesting' results. if you've spoken with sm in the past you've possibly already covered what i'm alluding to... something scott elsewhere appears to have successfully implemented and which i previously was unable. hint : ratchet.

if his results weren't bogus i believe this could be one instance of a ratchet.
Congratulations ScalaFX for coming out of the closet!
  • #27
  • Sep 9, 2012 3:43am Sep 9, 2012 3:43am
  •  regme
  • | Joined Mar 2010 | Status: Planning to be like Warren Buffett | 115 Posts
I use slightly different martingale multiplier lot multiplier, numbers are 1,2,2,2,4,4,4.... . When trade loss occur I revert back to 1.
  • #28
  • Sep 9, 2012 3:51am Sep 9, 2012 3:51am
You can add funds to the deposit after some period of drowdown and trade with the same risk as before (as persentage of the balance), but now as you deposit is bigger you will recover faster when the winning streak occur
  • #29
  • Sep 10, 2012 9:32am Sep 10, 2012 9:32am
Quoting lolik2020
Hi,
suppose you have an ea with win/lose ratio above 50%
and high expectency
is it Better 2 increase lot size after win or after loss?
i do not mean double lot size like martingale
it is very risky,
i mean increasing it by 20% or so
please vote above ^^^^^

I dont think so, its good decision to increase the lot size after getting loss/profit. The good decision is to keep on the winning track. Just concentrate on the strategy which is useful all the time. After getting profit I used to wait for the right time to open the position with the help of indicators, and never get rush for lots.
  • #30
  • Sep 22, 2012 6:53am Sep 22, 2012 6:53am
Quoting smjones
I created a coin toss indicator that has a margin of error of less than 1/100,000 of 1 percent randomness in 100 million coin tosses. (See screen shot)

By taking the Balance and dividing it in half and only using half of the balance to risk, and then setting the bet size as 100 percent of the half to risk, and by placing a full buy bet of 1 bet if the previous coin toss was heads, or placing a 1/2 buy bet of 0.5 if the previous coin toss was tails AND placing a limit of >= 200 for a max profit or a max limit of >= 25 for...

Thanks for this post, point 8 is key, bin thinking bout it for years.
  • #31
  • Sep 22, 2012 2:51pm Sep 22, 2012 2:51pm
Please come into my thread as the link given below. I have posted many examples with multiple solution with proven.

http://www.forexfactory.com/showthread.php?t=384691
  • #32
  • Nov 15, 2013 4:40am Nov 15, 2013 4:40am
The logic being this: in a 'perfect' 50/50 random walk, every trial is independent, hence every HH or TT sequence is equally as probable as a HT or TH sequence

probabilistics intepretation of toss and heads game and probabilistics intepretation asset pricing is different thing.

Simple example : it is possible get 15000 heads in row in casino ? Yes, it is. But it is possible that EURUSD is going down every 1 munute in row ? No - it is not because there is finite space - price can not be less than zero!

The key to crack the forex market using martingale-style strategy is:
1. MM
2. Assets selection and composition into one portfolio

I can agree, asset price of single asset is random walk
  • #33
  • Aug 18, 2014 2:18pm Aug 18, 2014 2:18pm
Quoting TheMaxx
Always risk a fixed percentage of capital on each trade. So increase your lot size after a win and decrease your lot size after a loss -> risking a fixed % each trade
agreed. you need to earn your right to trade big lots.
  • #34
  • Aug 18, 2014 2:20pm Aug 18, 2014 2:20pm
Quoting Mingary
1) trade small 2) no SL Scenario: you have 2 open trades: 3) lucky you you have winner a) keep it, close it, add to it ... it's a winner you are making some $$ b) do it again 4) back luck your other trade is now down 500 pips a) Use every reversal to reduce position. b) Lucky you, you waited it out and you bought at a reversal. Up 100 pips ==> close for 100 pips profit and close a 100 pips loss = loss 400 pips. Start again at 3)

are you nuts or what ?
  • #35
  • Aug 18, 2014 10:16pm Aug 18, 2014 10:16pm
In my experience you will never be successful changing your lot sizes from one win or loss to the next
  • #36
  • Aug 18, 2014 10:32pm Aug 18, 2014 10:32pm
Position size should be in keeping with probabilities on offer, i.e. higher probability setup justifies higher pos size. Extrapolating, that means trading with non-zero pos size when a suitable setup appears; and not trading (effectively zero pos size), when your rules/knowledge tell you otherwise.

If you can prove serial correlation — i.e. win is more likely, than not, to follow a previous win — over a large enough sample size, then that might also vindicate increasing pos size following a winning trade.

Otherwise, you're just playing a lucky guessing game of hit or miss.
  • #37
  • Aug 19, 2014 12:02pm Aug 19, 2014 12:02pm
Good trades/Good setups fail. If you usually trade 1 contract and decide to go with 2 because it's a high probability trade and you lose then what? You continue to trade with 1 contract? Or stay with 2 and lose more? It's 2 steps forward 3 steps back.

Consistency is what makes a good trader.
  • #38
  • Dec 12, 2017 3:14am Dec 12, 2017 3:14am
This is old thread. But anyway I will answer it based on what happened yesterday while I was trading.
I double the size in 1,2,3,4,5 sequence.
But this has the tendency to affect psychology.
I.e. You will be too engrossed in fighting the same market and not go away do another market instead.
At the end, I end up with size 10, and I have no ability to cut loss, because if I do, I will confirm a loss right away.
End up I trade till late and couldn't sleep over it.

How a size 1 trade ballooned to size 10 trade is truly beyond me...
I have to wake up my idea. :/
  • #39
  • Dec 12, 2017 3:59am Dec 12, 2017 3:59am
I think lot size should depend on balance amount like symphony system i found on FF.follow these strict rules
Attached File(s)
File Type: xls Symphonie System Weekly Growth Plan -5%.xls   76 KB | 608 downloads
  • #40
  • Last Post: Jan 27, 2018 6:30am Jan 27, 2018 6:30am
Quoting lolik2020
Hi, suppose you have an ea with win/lose ratio above 50% and high expectency is it Better 2 increase lot size after win or after loss? i do not mean double lot size like martingale it is very risky, i mean increasing it by 20% or so please vote above ^^^^^
No it's a bad idea to increase lot if you win a trade or loss. You can increase lot of your capital increase in good number. Suppose your capital is 5000$ and you take .05 lot. You can make it 0.06 once your capital is 6000$. You have to keep patience.
0 traders viewing now