Price Prediction
Hi daet,
I am not sure that I understood but you can look at this thread:
You have a bold claim there. FWIW, here's something to start with. To work with your calculation, start defining 'accuracy' in two axis: time and price.
- let's say your calculation calls out a price level (y) within the next time interval (x days max, for example). Three things can happen: target is met before stoploss is hit, stoploss is hit before target is hit, or none are hit but time is expired. Each event will require a decision.
- for each time/bar interval, make a prediction and study if the predicted level gets hit within the specified time horizon and how much drawdown is encountered before it hits
- gather all those data for at least 100 data points and create a distribution study to create a bail out level when drawdown is abnormally too large, and how frequent it happens. Why? Because you cannot martingale forever.
- compare the distance from entry to target vs entry to bailout level and the probability/frequency of each event to gather two numbers: win probability and risk-adjusted return then feed it into Kelly formula. You will be looking for a positive value closer to 1.
Highly doubt it!
I highly highly doubt with 100% accuracy that your little system can predict where that pair will be in 1-6 days with 100% accuracy. It makes no sense. You cannot predict where the market will be with 99% accuracy. You might be able to now, but it wont last and will drop to less than 50% soon unless you have a super computer supporting a neural network that is computing millions of possibility every second, and even that I do not have much confidence in since it is just re-optimizing every time.
And I hope you know with that 1% you wont be able to predict the market, is going to destroy your account with a trend that will never return. By the time this happens, your system would have tricked you so far into thinking it is reliable and that the market will return, but this one time it does not, is all you need to wipe your account out and traumatize you forever.
James, your condescending mail adds little value but thanks anyway.
Jonfraun you make a good point and i will start collecting data. I usually do not advocate martingale but a directional martingale could work. Another option would be selling options on the the pair as this gives you some wiggle room before closeout. For example, i know with a fair certainty that the EUR/USD will hit 1.4500 either today or by week end, so i started buying on the 28th..my account is up around $9K this morning (GMT+1). This is a $70K demo account which i traded up from $10K. As James so succinctly pointed out, its always that 1% that worries me so where do you place stops, this system needs big stops OR try and predict where the market will turnaround. Watch that 4500 level... 
You are right, I am apologize for having a less than helpful tone. It irritates me sometimes how new traders believe that trading is easy and a peace of cake as soon as they get two winning trades in a row and boast. I don't think you are a new trader and you know more or less what you are doing, so I take back my criticism. Perhaps you can average out the draw down levels, and if its around 500 pips for example, then you can set a hard stop at 500 pips so that trade that never comes back wont matter much.
Well today was classic, the EUR strengthened then fell out of bed. I guess this is on the back of the last minute US debt bailout. Fortunately i had locked it some decent profits with SL. Ive reopened some long positions for the predicted march to 4500. Now here a directional martingale would have you $^^$% your pants, i'm running a martingale strat on a VPS server and its sweating bullets which really confirms my gut feeling that martingale will NEVER be profitable long term.
Well today was classic, the EUR strengthened then fell out of bed. I guess this is on the back of the last minute US debt bailout. Fortunately i had locked it some decent profits with SL. Ive reopened some long positions for the predicted march to 4500. Now here a directional martingale would have you $^^$% your pants, i'm running a martingale strat on a VPS server and its sweating bullets which really confirms my gut feeling that martingale will NEVER be profitable long term.
RE - Martingale............. imho... very bad idea... various ideas and strategies is what makes a market... but any form of martingale is not for me...
Martingale only works if your trading strategy has over a 50% win rate.
Martingale only works if your trading strategy has over a 50% win rate.
James, you are simplifying too much... I can demonstrate a strategy with a very high win rate yet monetarily it is a losing strategy... Conversely, I can demonstrate a strategy that wins less than 50% of the time but monetarily is a winning strategy. I know you understand this simply from the nature of your past posts. Money management is everything and minimizing losses is everything and letting winners run is everything. I personally had a strategy trading es that I used back in 2007... I had 98 winners in a row going for a tp of 1 es point in one 30 day period... however at that time I utilized a 4 pt stoploss.... My win to loss ratio was way above 50%... however, it did not take many losers in a row compounded by martingale to wipe away my profits. Now, I am the first to admit that this strategy is flawed. However, I see strategies daily which employ a stoploss that would easily wipe out 4 winners... This is simply my experience... but a strategy which uses a tp that is smaller than its stoploss is potentially destined for failure. If you compound this with martingale... your devastation will simply happen sooner. Again, this is simply imho.
Regards
TCT
The biggest misconception about speculation is to be fixated with ONE of the TWO primary numbers. They are win probability and reward ratio (to risk). You will need both. There is nothing wrong with inverted reward ratio (reward / risk < 1) so long as: the strategy is robust and the probability of win confirms that it is not too close to the breakeven number. Why not too close? Because probability is not precise and it will be precise along with higher number of samples just like if you flip a coin 10 times you will have larger variance from ending up with 50% heads and 50% tails than doing it 1000 times.
The problem is the spread once adjusted into the reward/risk ratio. If you trade small movement magnitude closer to spread size, inverted reward ratio will have bigger issue as winning will become significantly much smaller and easily tilt the ratio into negative Kelly number. A technical issue, but one that needs extra attention and may carry the final word.
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
You agree to website policy and terms of use
Ive worked out a system which tells me with 99% accuracy where the EUR/USD will be at a certain time in the future. Using this system ive doubled my account in 15 days. The problem is however that it can take anything from 1 day to 5-6 days to reach the predicted level and you have to withstand drawdown until the market reverses. I'm looking to develop s system that enables me to trade this particular scenario. For example, a directional martingale would work as your entry does not have to be that accurate and you can scale up using mini or micro lots. Any ideas would be appreciated. I'm prepared to share my system with anyone who comes up with the winning system.