Robot Trading on futures - ES et al
- Thread starter iuykcif
- Start date
There is also one intuitive reason why you can't do in 1 account.
You are scalping in 1 account.
Imagine you make an entry LONG at price P buying a quantity Q.
If the price goes up P + dP you close and have scalped some amount L *Q * dP - (expenses) $
If the price goes down you begin to lose.
All you lose is lost, and to begin to see some gain you have to - first - return to a level > P (due to spread + commissions).
The beauty of this particular algorithm is that whatever price movement in any direction ** will
necessarily (and not probabilistically) ** be transformed into profit, whenever there is a reversal (before expiry clearly).
The ** only condition ** is that you have enough fund to hold the oscillation (or adopted the necessary
measures and precautions to be within the margins). For the rest, the whole oscillation, will become profit.
<b>There is no oscillation (provided it's bigger than the ScalpRange) that you can hold, which will
not turn into profit, if it reverses before expiry. This is guaranteed by design, and it's not a probabilistic matter.</b>
The logic is that ** every piece ** of the price curve (larger than the prespecified ScalpRange) - in any direction - will eventually be transformed, on reversion,
into profits.
You are scalping in 1 account.
Imagine you make an entry LONG at price P buying a quantity Q.
If the price goes up P + dP you close and have scalped some amount L *Q * dP - (expenses) $
If the price goes down you begin to lose.
All you lose is lost, and to begin to see some gain you have to - first - return to a level > P (due to spread + commissions).
The beauty of this particular algorithm is that whatever price movement in any direction ** will
necessarily (and not probabilistically) ** be transformed into profit, whenever there is a reversal (before expiry clearly).
The ** only condition ** is that you have enough fund to hold the oscillation (or adopted the necessary
measures and precautions to be within the margins). For the rest, the whole oscillation, will become profit.
<b>There is no oscillation (provided it's bigger than the ScalpRange) that you can hold, which will
not turn into profit, if it reverses before expiry. This is guaranteed by design, and it's not a probabilistic matter.</b>
The logic is that ** every piece ** of the price curve (larger than the prespecified ScalpRange) - in any direction - will eventually be transformed, on reversion,
into profits.
Hi iuykcif,
'Whenever there is a reversal' that is the underlying premise of the strategy, that there is always a reversal of some sort.
I was wondering if, through your programming, you can come up with some statistical probabilities? For example, how many days over a 100 day period did price return to a level reached, say 2 hours earlier in the day?
i.e. Price hit 1000.50 at 10:05 am and then goes up to 1008. How many times on average does it return to 1000.50? (using each individual day's price range, of course).
Basically I'm trying to see what prevents the system from averaging into oblivion, if by black swan chance, the market runs away and never comes back. It's unlikely, I know, but you have to consider it, as risk manager is part of the trader profile
I didn't realize the similarity, but a month ago I thought of a similar (similar in premise only, definitely not in profitability, sophistication or time spent lol) scalping/average in system , but being so n00bular in my thought process, I failed to understand nothing was stopping me from averaging into oblivion unless I used a fixed stop.
If you could assign a statistical probability that price returns to a previous price x% of times each day, I think that would give you and potential users of the system even greater confidence than the supreme backtests?
BTW, did you do any backtests over September/October of last year in the very volatile markets? I think the drawdowns from that period would be the best indicator if the system has longer term viability, being able to survive those tumultuous times.
Now with my observations out of the way, I'll try to make an enhancement Of course, you already discussed being long premium to offset unrealized losses in a runaway market.
Another thing, and I think you probably don't use indicators, but it's worth a shot:
Only running the system based on the rate of change of 'Average True Range' Indicator.
Run robot IF Rate of Change of ATR < +x
The premise is that if the ATR is not expanding, the market is likely staying within a range in which the robot can scalp profitably. Once ROC of ATR > x, the robot rebalances and doesn't scalp until the rate of change starts to decline, meaning the market is losing one-way momentum and may become viable to scalp again.
If you did this in conjunction with long straddles, all the better. With expanding ATR, you have a better probability of profiting from your long premium, and you don't have to worry about scalping. Then, once the ATR starts to detract, you could offset your long premium (perhaps at a profit), buy a new straddle, and run the robot again.
Quick idea, maybe using $TICK extremes. Just throwing it out there, don't utilize TICK myself so I'm not sure how to implement.
Please flame me for wanting to use an indicator lol
'Whenever there is a reversal' that is the underlying premise of the strategy, that there is always a reversal of some sort.
I was wondering if, through your programming, you can come up with some statistical probabilities? For example, how many days over a 100 day period did price return to a level reached, say 2 hours earlier in the day?
i.e. Price hit 1000.50 at 10:05 am and then goes up to 1008. How many times on average does it return to 1000.50? (using each individual day's price range, of course).
Basically I'm trying to see what prevents the system from averaging into oblivion, if by black swan chance, the market runs away and never comes back. It's unlikely, I know, but you have to consider it, as risk manager is part of the trader profile
I didn't realize the similarity, but a month ago I thought of a similar (similar in premise only, definitely not in profitability, sophistication or time spent lol) scalping/average in system , but being so n00bular in my thought process, I failed to understand nothing was stopping me from averaging into oblivion unless I used a fixed stop.
If you could assign a statistical probability that price returns to a previous price x% of times each day, I think that would give you and potential users of the system even greater confidence than the supreme backtests?
BTW, did you do any backtests over September/October of last year in the very volatile markets? I think the drawdowns from that period would be the best indicator if the system has longer term viability, being able to survive those tumultuous times.
Now with my observations out of the way, I'll try to make an enhancement
Of course, you already discussed being long premium to offset unrealized losses in a runaway market. Another thing, and I think you probably don't use indicators, but it's worth a shot:
Only running the system based on the rate of change of 'Average True Range' Indicator.
Run robot IF Rate of Change of ATR < +x
The premise is that if the ATR is not expanding, the market is likely staying within a range in which the robot can scalp profitably. Once ROC of ATR > x, the robot rebalances and doesn't scalp until the rate of change starts to decline, meaning the market is losing one-way momentum and may become viable to scalp again.
If you did this in conjunction with long straddles, all the better. With expanding ATR, you have a better probability of profiting from your long premium, and you don't have to worry about scalping. Then, once the ATR starts to detract, you could offset your long premium (perhaps at a profit), buy a new straddle, and run the robot again.
Quick idea, maybe using $TICK extremes. Just throwing it out there, don't utilize TICK myself so I'm not sure how to implement.
Please flame me for wanting to use an indicator lol
Further to the previous post and some of your own comments regarding TA, I really can't see how you can suggest TA in some form or another doesn't work when your algorithm itself must use a trigger of some sort to enter trades.
Looking at your last screenshot its quite apparent that something is working away making a decision when to enter a trade or not, even if its something as simple as a change in direction of the price.
That is using TA is it not ?
I quite understand we're talking semantics here but you're the one who has stated TA doesn't work...
If I'm missing something obvious please let me know as I'm always happy to be corrected..............really.
Looking at your last screenshot its quite apparent that something is working away making a decision when to enter a trade or not, even if its something as simple as a change in direction of the price.
That is using TA is it not ?
I quite understand we're talking semantics here but you're the one who has stated TA doesn't work...
If I'm missing something obvious please let me know as I'm always happy to be corrected..............really.
I think the robot's entry/rebalance is just on price action once movement exceeds the predefined ScalpRange, like 1 point or something. If it is constrained to a two contract spread, it won't rebalance until price moves to a profitable position. If unconstrained, it will keep averaging in.
still doubt that this strategy has no risk as OP stated. it is basically an averaging-down strategy. the two accounts only make it more deceiving. if there are no constraints, it is idential to normal averaging down. with constraints, it is similar to the cut loss of averaging down. any one can prove it is not averaging down?
_
Thanks Kedwards, for your kind comments.
<b>--Now with my observations out of the way, I'll try to make an enhancement Of course, you already discussed being long premium to offset unrealized losses in a runaway market.
Another thing, and I think you probably don't use indicators, but it's worth a shot:
Only running the system based on the rate of change of 'Average True Range' Indicator.
Run robot IF Rate of Change of ATR < +x
The premise is that if the ATR is not expanding, the market is likely staying within a range in which the robot can scalp profitably. Once ROC of ATR > x, the robot rebalances
and doesn't scalp until the rate of change starts to decline, meaning the market is losing one-way momentum and may become viable to scalp again.
If you did this in conjunction with long straddles, all the better. With expanding ATR, you have a better probability of profiting from your long premium, and you don't have to
worry about scalping. Then, once the ATR starts to detract, you could offset your long premium (perhaps at a profit), buy a new straddle, and run the robot again. </b>
You are perfectly right. That's what I have been looking into in these days to find a solution if you are not a <b>fund</b>, and you do not have huge funds on your accounts to make this market making program run and collect money all the way, and through all possible oscillations.
The idea was, as you mention, to place one (or more) straddle(s) with enough options so that if the price breaks out we will possibly recover and profit from the straddle and if not we will be scalping inside the straddle. At the same time when the price is near or past the strikes you open new independent instances of the robot with another straddle, and so on.
This way you will be scalping forever and usually don't get wrong more than 1 straddle.
So always new independent robot instances at difference price levels, all protected by straddles. This seemed an obvious scenario for the medium investor (small investors and retail traders stay away from futures!.
Frankly if i could find a different solution than options I would be happier, and if one has scarce money I would only suggest to try relatively *very* small packets on ETF, avoiding options. Anyone can trade, relatively speaking, "like a sheik" if he trades amounts *very* small relative to its capital. It's all a question of proportions.
You have little money? You certainly don't go futures unless you want to entertain yourself with your own liquidation. Make small packets and be patient: you will have your little reward.
It's just the oscillating nature of prices which doesn't allow (statistically speaking) making money from little money. You have in fact able to hold on one side of the oscillation to be profitable.
<b>--Quick idea, maybe using $TICK extremes. Just throwing it out there, don't utilize TICK myself so I'm not sure how to implement.
Please flame me for wanting to use an indicator lol </b>
How could I? Being from Rome, I descend from people who believed the gods sent signs of future events to men through the flight of birds, or divinate observing a cat's manner of jumping or other natural phenomena. It's perhaps because of that inheritance that I have come to the opposite end of holding that indicators are useful only if you need a
signal to understand when you are about starting losing money) (flames expected)
No, currently I favor "deterministic" approaches. Had too much probability in my life to trust it, when money is concerned)
This particular algorithm (which is indeed highly complex and difficult to implement, although in my first replies I jokingly gave hints that was "just" scalping) offer a "deterministic" guarantee about the fact that it will turn any oscillation into profit, it's not based on "maybe" or probability. It just does it.
A "by design" guarantee that you will hardly find with other approaches.
TOm
Thanks Kedwards, for your kind comments.
<b>--Now with my observations out of the way, I'll try to make an enhancement Of course, you already discussed being long premium to offset unrealized losses in a runaway market.
Another thing, and I think you probably don't use indicators, but it's worth a shot:
Only running the system based on the rate of change of 'Average True Range' Indicator.
Run robot IF Rate of Change of ATR < +x
The premise is that if the ATR is not expanding, the market is likely staying within a range in which the robot can scalp profitably. Once ROC of ATR > x, the robot rebalances
and doesn't scalp until the rate of change starts to decline, meaning the market is losing one-way momentum and may become viable to scalp again.
If you did this in conjunction with long straddles, all the better. With expanding ATR, you have a better probability of profiting from your long premium, and you don't have to
worry about scalping. Then, once the ATR starts to detract, you could offset your long premium (perhaps at a profit), buy a new straddle, and run the robot again. </b>
You are perfectly right. That's what I have been looking into in these days to find a solution if you are not a <b>fund</b>, and you do not have huge funds on your accounts to make this market making program run and collect money all the way, and through all possible oscillations.
The idea was, as you mention, to place one (or more) straddle(s) with enough options so that if the price breaks out we will possibly recover and profit from the straddle and if not we will be scalping inside the straddle. At the same time when the price is near or past the strikes you open new independent instances of the robot with another straddle, and so on.
This way you will be scalping forever and usually don't get wrong more than 1 straddle.
So always new independent robot instances at difference price levels, all protected by straddles. This seemed an obvious scenario for the medium investor (small investors and retail traders stay away from futures!
.Frankly if i could find a different solution than options I would be happier, and if one has scarce money I would only suggest to try relatively *very* small packets on ETF, avoiding options. Anyone can trade, relatively speaking, "like a sheik" if he trades amounts *very* small relative to its capital. It's all a question of proportions.
You have little money? You certainly don't go futures unless you want to entertain yourself with your own liquidation. Make small packets and be patient: you will have your little reward.
It's just the oscillating nature of prices which doesn't allow (statistically speaking) making money from little money. You have in fact able to hold on one side of the oscillation to be profitable.
<b>--Quick idea, maybe using $TICK extremes. Just throwing it out there, don't utilize TICK myself so I'm not sure how to implement.
Please flame me for wanting to use an indicator lol </b>
How could I? Being from Rome, I descend from people who believed the gods sent signs of future events to men through the flight of birds, or divinate observing a cat's manner of jumping or other natural phenomena. It's perhaps because of that inheritance that I have come to the opposite end of holding that indicators are useful only if you need a
signal to understand when you are about starting losing money
) (flames expected)No, currently I favor "deterministic" approaches. Had too much probability in my life to trust it, when money is concerned
)This particular algorithm (which is indeed highly complex and difficult to implement, although in my first replies I jokingly gave hints that was "just" scalping) offer a "deterministic" guarantee about the fact that it will turn any oscillation into profit, it's not based on "maybe" or probability. It just does it.
A "by design" guarantee that you will hardly find with other approaches.
TOm
As far as I can tell (and I may be wrong here) this is just another averaging system. The use of 2 accounts is handy from an execution perspective but it suffers from the same downfall as all averaging strategies...you need to have enough dosh to keep averaging until the price returns to the mean. 99.99% of the time this will be achievable but once in a while a big black swan will swim along and bite you hard in the arse whilst you keep on averaging. These kinds of strategies are the most seductive (and thus most dangerous) because they produce such great backtesting results.
I would like to see how this strategy copes
in a limit locked market or generally in liquidity vacuums.
I would like to see how this strategy copes
in a limit locked market or generally in liquidity vacuums.