Back-tested results show 20k to 200k in 9 months !

[Narass]

Hi all

Recently i designed a very simple strategy for Pure day trading.

I set out to back-test this strategy using VB, i was amazed to find the results. They are given below.

When Compounded
LONG SL Hits=31 PROFIT HITS=92 Not Triggered=36 NO RESULT(time)=11 NO RESULT(5-loops)=47
SHORT SL Hits=45 PROFIT HITS=101 Not Triggered=33 NO RESULT(time)=9 NO RESULT(5-loops)=35
End Capital=206476
Total Net Profit Compounded=186476


When not compounded, keeping Cap as 20k throughout
LONG SL Hits=31 PROFIT HITS=92 Not Triggered=36 NO RESULT(time)=11 NO RESULT(5-loops)=47
SHORT SL Hits=45 PROFIT HITS=101 Not Triggered=33 NO RESULT(time)=9 NO RESULT(5-loops)=35
End Capital=20000
Total Net Profit=49354



As i didnt use Ami for backtesting, iam not sure how to evaluate this strategy further. I was also advised that Net profit is not a very good measure.

Some points:

>> This is purely intraday strategy
>> R:R = 1:1, so really needed a good win% to make it successful
>> Tested only on Liq Stocks and not on index
>> Tested on data from May2010 to till date.
>> Trades started with Cap of 20k.
>> Qty traded is for the entire capital on hand.
>> Strategy continues from yesterday, meaning it takes candles from yday for trading on the first bar of today.
>> SL or Profit hits are evaluated till 3:15pm
>> Profit figure given above is net of brokerage+taxes on all executed trades.
>> There is no clarity as of now for those trades where it has not hit the Tgt or SL in the next 5 candles. I have kept it separate. The above profit figure takes those trades as "Exit at Entry" cases and brokerage deducted.
>> Similar treatment for "No Result (time)" where it cud not close a trade till 3:15 pm.
>> Tested on some of other scrips, generally gives fav results
>> Generally checked the accuracy of the backtester code and seems to be good till now.


What is the way forward.
What else measures to test/check
Is this tradeable
What time period to backtest
Some tips on Forward testing
Other left out points

 

[sumosanammain]

Nice results......

Will forward my loan application after 9 months to you

 

[Narass]

Tested another scrip and the results are given below. Here again the initial cap is 20k.

When Profit is compounded
Long SL Hits=38 PROFIT HITS=103 Not Triggered=36 NO RESULT(time)=7 NO RESULT(5-loops)=32
Sell SL Hits=52 PROFIT HITS=99 Not Triggered=37 NO RESULT(time)=7 NO RESULT(5-loops)=31
TOTAL NET PROFIT=163191


When not compounded
Long SL Hits=38 PROFIT HITS=103 Not Triggered=36 NO RESULT(time)=7 NO RESULT(5-loops)=32
Sell SL Hits=52 PROFIT HITS=99 Not Triggered=37 NO RESULT(time)=7 NO RESULT(5-loops)=31
TOTAL NET PROFIT=47086

 

[sumosanammain]

Tested another scrip and the results are given below. Here again the initial cap is 20k.

When Profit is compounded
Long SL Hits=38 PROFIT HITS=103 Not Triggered=36 NO RESULT(time)=7 NO RESULT(5-loops)=32
Sell SL Hits=52 PROFIT HITS=99 Not Triggered=37 NO RESULT(time)=7 NO RESULT(5-loops)=31
TOTAL NET PROFIT=163191


When not compounded
Long SL Hits=38 PROFIT HITS=103 Not Triggered=36 NO RESULT(time)=7 NO RESULT(5-loops)=32
Sell SL Hits=52 PROFIT HITS=99 Not Triggered=37 NO RESULT(time)=7 NO RESULT(5-loops)=31
TOTAL NET PROFIT=47086

Watch out, you may be approached by top mutual funds for managing their billions!
I have gone through this phase in 2002 and can identify with your statements.
It took 7 long years to realize that its not how it works.

 

[linkon7]

Hi all

Recently i designed a very simple strategy for Pure day trading.

I set out to back-test this strategy using VB, i was amazed to find the results. They are given below.

When Compounded
LONG SL Hits=31 PROFIT HITS=92 Not Triggered=36 NO RESULT(time)=11 NO RESULT(5-loops)=47
SHORT SL Hits=45 PROFIT HITS=101 Not Triggered=33 NO RESULT(time)=9 NO RESULT(5-loops)=35
End Capital=206476
Total Net Profit Compounded=186476


When not compounded, keeping Cap as 20k throughout
LONG SL Hits=31 PROFIT HITS=92 Not Triggered=36 NO RESULT(time)=11 NO RESULT(5-loops)=47
SHORT SL Hits=45 PROFIT HITS=101 Not Triggered=33 NO RESULT(time)=9 NO RESULT(5-loops)=35
End Capital=20000
Total Net Profit=49354



As i didnt use Ami for backtesting, iam not sure how to evaluate this strategy further. I was also advised that Net profit is not a very good measure.

Some points:

>> This is purely intraday strategy
>> R:R = 1:1, so really needed a good win% to make it successful
>> Tested only on Liq Stocks and not on index
>> Tested on data from May2010 to till date.
>> Trades started with Cap of 20k.
>> Qty traded is for the entire capital on hand.
>> Strategy continues from yesterday, meaning it takes candles from yday for trading on the first bar of today.
>> SL or Profit hits are evaluated till 3:15pm
>> Profit figure given above is net of brokerage+taxes on all executed trades.
>> There is no clarity as of now for those trades where it has not hit the Tgt or SL in the next 5 candles. I have kept it separate. The above profit figure takes those trades as "Exit at Entry" cases and brokerage deducted.
>> Similar treatment for "No Result (time)" where it cud not close a trade till 3:15 pm.
>> Tested on some of other scrips, generally gives fav results
>> Generally checked the accuracy of the backtester code and seems to be good till now.


What is the way forward.
What else measures to test/check
Is this tradeable
What time period to backtest
Some tips on Forward testing
Other left out points

back test results dont take into account the human emotion factor. My favorite method of test a system is to use bar replay and taking notes on the trades on an excel sheet. This also gives a confidence put the system into a live test with smaller quantity. After every week... note the performance and Grey areas and work on them. Once satisfied then gradually put more money at risk...

This process is time consuming but worth the effort as it gives u confidence to put your hard earned money at risk...

 

[Raju]

Linkon7,

I need help regards barreplay.In Amibroker when I try to replay for 1 min it does not work.Any specific settings need to be done ?

Regards,
Raju

 

[linkon7]

Linkon7,

I need help regards barreplay.In Amibroker when I try to replay for 1 min it does not work.Any specific settings need to be done ?

Regards,
Raju

if u are having a IEOD data with atleast 1 min tf as your base interval...then u should be able to see bar replay without any problem.

 

[Smart_trade]

Tested another scrip and the results are given below. Here again the initial cap is 20k.

When Profit is compounded
Long SL Hits=38 PROFIT HITS=103 Not Triggered=36 NO RESULT(time)=7 NO RESULT(5-loops)=32
Sell SL Hits=52 PROFIT HITS=99 Not Triggered=37 NO RESULT(time)=7 NO RESULT(5-loops)=31
TOTAL NET PROFIT=163191


When not compounded
Long SL Hits=38 PROFIT HITS=103 Not Triggered=36 NO RESULT(time)=7 NO RESULT(5-loops)=32
Sell SL Hits=52 PROFIT HITS=99 Not Triggered=37 NO RESULT(time)=7 NO RESULT(5-loops)=31
TOTAL NET PROFIT=47086

Seems to be a workable strategy. My observations are as under :

1) Profit factor of 1:1 is a bit low....this means we need larger percentage of successful trades to remain on the positive side of our ledger.

2) Number of consecutive losses and the maximum and average amount of drawdowns not known....

3) Any provision of adding in profitable trades ? Will it improve the avarage points made / average points lost ratio ?

4) What is the period on which this strategy was tested ? No of days and does it include trending as well as choppy periods ?

Best wishes for a winning strategy....

Smart_trade

 

[Narass]

This is one of other resuts which i wud like to share

When keeping the Cap as constant 20k (making up when in loss). The only change here is that, I exit at close of 5th candle in case of "No Result(5-loop)".
Long SL Hits=57 PROFIT HITS=151 Not Triggered=45 NO RESULT(time)=11 NO RESULT(5-loops)=44
Sell SL Hits=76 PROFIT HITS=231 Not Triggered=49 NO RESULT(time)=10 NO RESULT(5-loops)=40
TOTAL NET PROFIT= 69548

When Compounded (ploughing the capital back/dont replenish losses/Exit at 5th candle close when no result)
Long SL Hits=57 PROFIT HITS=151 Not Triggered=45 NO RESULT(time)=11 NO RESULT(5-loops)=44
Sell SL Hits=76 PROFIT HITS=231 Not Triggered=49 NO RESULT(time)=10 NO RESULT(5-loops)=40
TOTAL NET PROFIT= 622525

The compounding may not work as towards the end of the testing period, the capital grows higher and higher and becomes no longer viable to trade in cash mkt.

 

[Narass]

Thanx for replying and my replies are given below ur questions

Seems to be a workable strategy. My observations are as under :

1) Profit factor of 1:1 is a bit low....this means we need larger percentage of successful trades to remain on the positive side of our ledger.

--- yup been trying to work on >1:1, but results are awry. I understand thru my watching charts, that there are certain periods of time during the day, probability of getting >1:1 seems good. I can tinker that idea a bit and try to code it in.

2) Number of consecutive losses and the maximum and average amount of drawdowns not known....

--- i have not calculated those. Let me do it

3) Any provision of adding in profitable trades ? Will it improve the avarage points made / average points lost ratio ?

--- i have not done that. Working on avg points won and lost and its ratio

4) What is the period on which this strategy was tested ? No of days and does it include trending as well as choppy periods ?

--- I have backtested from May2010 to till date. I noticed that there were many days in May10, which were in losses. Getting data is another issue, esp intraday data.