Conservative strategy

marimuthu13

Well-Known Member
@all, I was curious to see how this strategy behaves for a longer time and wrote a program to backtest on BNF. Contrary to my belief, the strategy
produced exceptional results.

Following are the slightly modified rules I considered:
1. Use SupterTrend settings as suggested by Suri.
2. Use the entries as suggested by Suri.
3. When the stop is hit for the firstTrade, enter the next trade with (2 * lots) of the firstTrade. If the stop is hit again in this
trade, enter the next trade with (3 * lots). If the stop is hit again restart with the original lot size (1 * lots).
3. Do not enter when a trade is triggered in the same bar when a previous trade was closed.
4. No break in trades as suggested.

Following is the result:
Backtest Data: Jan 2011 to July 2016
Close positions at 3:20 PM.
Brokerage: 4 points.
StopLoss: 25, Target: 75
Number of Trades: 3827
Total Points (after brokerage): 47489
Monthly average number of trades: 56
Monthly average number of times stop hit: 33
Number of Losing months: 7
Number of Winning months: 61
Longest Losing streak: 18
Losing streak Drawdown: -960.0
Longest Winning streak: 8

Result for NF:
Brokerage: 2 points.
StopLoss: 10, Target: 30
Number of Trades: 3633
Total Points (after brokerage): 17051
Monthly average number of trades: 53
Monthly average number of times stop hit: 30
Number of Losing months: 12
Number of Winning months: 56
Longest Losing streak: 15
Losing streak Drawdown: -327.4
Longest Winning streak: 8

Do let me know if you need any other details or you think anything is incorrect above.
amazing results..

BNF average monthly points - 718
NF average monthly points - 260

can you please do back testing for crude oil also.
 

humble

Well-Known Member
On BNF the losing streak shows 18 trades i.e. 36 lots in all (1+2+3)*18/3
ie DD will be 36 x (25 + 4) = 1044

but your stats show 960 points, so things don't add up neatly

plz verify your code for some minor inconsistencies


Happy :)
Thanks for looking into it. Yes i did look into this inconsistency, which is why I updated my initial post to add "Close Position: 3:20 PM". On days where we hit EndOfDay condition i.e our trade is still in negative without hitting stoploss, the trade is closed at 3:20 PM. Let me know if it answers your concern.
 

lvgandhi

Well-Known Member
@all, I was curious to see how this strategy behaves for a longer time and wrote a program to backtest on BNF. Contrary to my belief, the strategy
produced exceptional results.

Following are the slightly modified rules I considered:
1. Use SupterTrend settings as suggested by Suri.
2. Use the entries as suggested by Suri.
3. When the stop is hit for the firstTrade, enter the next trade with (2 * lots) of the firstTrade. If the stop is hit again in this
trade, enter the next trade with (3 * lots). If the stop is hit again restart with the original lot size (1 * lots).
3. Do not enter when a trade is triggered in the same bar when a previous trade was closed.
4. No break in trades as suggested.

Following is the result:
Backtest Data: Jan 2011 to July 2016
Close positions at 3:20 PM.
Brokerage: 4 points.
StopLoss: 25, Target: 75
Number of Trades: 3827
Total Points (after brokerage): 47489
Monthly average number of trades: 56
Monthly average number of times stop hit: 33
Number of Losing months: 7
Number of Winning months: 61
Longest Losing streak: 18
Losing streak Drawdown: -960.0
Longest Winning streak: 8

Result for NF:
Brokerage: 2 points.
StopLoss: 10, Target: 30
Number of Trades: 3633
Total Points (after brokerage): 17051
Monthly average number of trades: 53
Monthly average number of times stop hit: 30
Number of Losing months: 12
Number of Winning months: 56
Longest Losing streak: 15
Losing streak Drawdown: -327.4
Longest Winning streak: 8

Do let me know if you need any other details or you think anything is incorrect above.
Can You share the afl for back testing this strategy, Please?
 
The logic has been programmed in Java and not an afl.
Yes, makes sense now when the trade is cut depending on time, loss can be less than SL limits

Can you test for few conditions like

1) Something similar to Suri where we stop trading after n number of trades every day.

2) Stop trading after reaching a Loss limit for the day.

3) Stop trading after reaching a Profit limit for the day.

4) No new trades after specific time each day say 2:00 to 3:10/15 pm.

5) Effect of changing the cut-off/square time say between 2:45 to 3:25pm


Happy :)
 

humble

Well-Known Member
Please find the answers/doubts inline.

Yes, makes sense now when the trade is cut depending on time, loss can be less than SL limits

Can you test for few conditions like

1) Something similar to Suri where we stop trading after n number of trades every day.
>> I did not mention this earlier, but we will only take 3 stops per day and after that no trades for the day. This condition is already in place.

2) Stop trading after reaching a Loss limit for the day.
>> Since the stoploss is defined as 10 points for NF. Max stoploss in a day can be [10 + (2*10) + (3*10)] = 60.

3) Stop trading after reaching a Profit limit for the day.
>> Do you have any specific limit you want me to try?

4) No new trades after specific time each day say 2:00 to 3:10/15 pm.
>> Right now no new trades are taken after 3:00 PM. This condition is already in place.

5) Effect of changing the cut-off/square time say between 2:45 to 3:25pm
>> I tested for closing times from 3:05 to 3:25. The monthly average points varies only from -7 to +3 [at 3:10] points.

Happy :)
 
Last edited:
Please find the answers/doubts inline.
Thanks

Except for the first one i.e. stop at 3 trades per day all other are minor changes. If the system is robust. the difference for all the minor changes should be very small, 5-10-20 points. But when we add up few such small edges, can make a difference of over all up to 10-20%.

Anyway since you already have code in place was curious to know "what if" in similar situations

If you can run the code with condition stop trading after 2/4/5 trades per day instead of fixed 3 trades, do let us know the results.

Once again thanks for your efforts.


Happy :)
 

humble

Well-Known Member
Thanks

Except for the first one i.e. stop at 3 trades per day all other are minor changes. If the system is robust. the difference for all the minor changes should be very small, 5-10-20 points. But when we add up few such small edges, can make a difference of over all up to 10-20%.

Anyway since you already have code in place was curious to know "what if" in similar situations

If you can run the code with condition stop trading after 2/4/5 trades per day instead of fixed 3 trades, do let us know the results.

Once again thanks for your efforts.


Happy :)
I hope I understood your request correctly, if not please let me know here/PM.

I increased the number of trades per day to 5 and with different number of stoploss hits per day following is the result:

5 trades per day, 5 stoploss, Monthly average Points: 314
5 trades per day, 4 stoploss, Monthly average Points: 312
5 trades per day, 3 stoploss, Monthly average Points: 305
5 trades per day, 2 stoploss, Monthly average Points: 290

From the result, it looks like more trades is increasing the profitability even after considering the brokerage. It would be worth looking at this strategy in the 3min timeframe. Other noticeable fact was that with 2 stoploss the number of -ve months increased from 12 to 15.

Anyways all this is theoretical what matters is implementing it effectively trade after trade. Thanks to Suri for sharing such a wonderful method in the forum.
 
I hope I understood your request correctly, if not please let me know here/PM.

I increased the number of trades per day to 5 and with different number of stoploss hits per day following is the result:

5 trades per day, 5 stoploss, Monthly average Points: 314
5 trades per day, 4 stoploss, Monthly average Points: 312
5 trades per day, 3 stoploss, Monthly average Points: 305
5 trades per day, 2 stoploss, Monthly average Points: 290

From the result, it looks like more trades is increasing the profitability even after considering the brokerage. It would be worth looking at this strategy in the 3min timeframe. Other noticeable fact was that with 2 stoploss the number of -ve months increased from 12 to 15.

Anyways all this is theoretical what matters is implementing it effectively trade after trade. Thanks to Suri for sharing such a wonderful method in the forum.
Nice :thumb:

Results seem on expected lines, more the trades more is the profit . . .

On a large base the probability distribution is random so somewhat equal probability of a winner after 1/2/3/4/5 losing streaks

When trading manually the rule of stopping after 3 trades will help preserve psychological capital :D

But if used with auto trading, it can be extended to 5T/5SL as your figuers suggest.

Thanks Suri for the thread, and Humble for the studies

Happy :)