Backtested Results - Spot On

[rog1111]

Since there seems to be quite a lot of discussion about the effectiveness of the Spot-On system, I have attached a link to an Excel file with backtested results for EUR:$.

Various hourly breakout scenarios are covered, and all figures have been generated by SureTracker software.

Please note the following :

1) The times are a little confusing - eg 06:59:59 refers to the 07h00 to 08h00 bar etc
2) These are Best scenario results. It's unlikely that you would have done better in practice. The 35 pip stop will be more affected by a best/worst assumption
3) When comparing results, the important figues are not only the Total PL, but the Max overall Drawdown and the PLD ratio. In other words, it's better to make 3000 pips with a max DD of -1000 pips rather than a DD of -3000 pips.

Results look good - here's the link :

click here

rog1111

 

[robber]

Great stuff Rog, I presume these calcs are based on pure data with no allowance made for spread or costs?

Neil

 

[rog1111]

Neil

That is correct, there is no allowance for slippage, commissions or spreads. Trade entry price is the exact high/low of the time range, if broken. Also 1 trade only per day, no reversal trades.

rog1111

Great stuff Rog, I presume these calcs are based on pure data with no allowance made for spread or costs?

Neil

 

[wellst2w]

Rog,
Thank you for providing back-test results. They are beyond my ability to produce.
I have two questions. First, should I calculate "No. of hours of entry" by subtracting "Start bar start time" from "End bar start time"+1? If so, my answers are different from yours. Am I in error?
Second, do the assumptions of any of your cases match that shown in 3rddawn's spreadsheet (post of 25-Mar-04, 1350 hrs)? 3rddawn's spreadsheet shows a point profit from 6 Jan to 24 Mar 2004 of 1419 pips. If there is a match, then does this means there have been losses in the last 6 weeks from 24 Mar to 7 May 2004 and the drawdown from 2 Jan to 7 May 2004 may be larger than you show?
Nick

 

[rog1111]

Nick

To take these 1 at a time.

"No of hours for Entry" means the number of hours allowed for entry after the initial price range as been established. IOW, if you are using the 7-10 system, and 2 hours are allowed for entry, you must have had a breakout and entry by 12, otherwise no trade after that.

Re 3rddawn's spreadsheet, I'll need to know exactly which system is being used. Do you have the link for the post ? If you talking about the standard JonnyT 7-10 system for EUR$, with a 35 pip stop, closing at 21h00 if not previously, and allowing entry anytime between 10h00 and 21h00, then for these dates inclusive, I get 1077 pips profit BEST . 25th March to 7th May = +26 pips (ex 24th march at +148, a big one)

It's worth noting that unless shorter term data is being used (10 min or less, ideally tick) to create hourly bars, results may be unreliable when working with tight(ish) stops (eg 35 pips), so the best we can do is to report best case, worst case & best guess results based on open & close. Wider stops are less likely to be affected.

Drawdown is as stated in the file.

rog1111

Rog,
Thank you for providing back-test results. They are beyond my ability to produce.
I have two questions. First, should I calculate "No. of hours of entry" by subtracting "Start bar start time" from "End bar start time"+1? If so, my answers are different from yours. Am I in error?
Second, do the assumptions of any of your cases match that shown in 3rddawn's spreadsheet (post of 25-Mar-04, 1350 hrs)? 3rddawn's spreadsheet shows a point profit from 6 Jan to 24 Mar 2004 of 1419 pips. If there is a match, then does this means there have been losses in the last 6 weeks from 24 Mar to 7 May 2004 and the drawdown from 2 Jan to 7 May 2004 may be larger than you show?
Nick

 

[Mr Chill]

Nice work rog1111 ,interesting results
Re SL ,0.006 does this equate to 6 points or 60 ?

 

[wellst2w]

Rog,
Many thanks for your explanations.
http://www.trade2win.co.uk/boards/showthread.php?t=8333&page=2&pp=40 is the website (look for post # 46) containing 3rddawn's email and spreadsheet attachment. 1 minute data was used.
3rddawn's results compare well from 6 Jan to 4 Mar 2004 with JonnyT's results using the March 2004 contract (see http://www.trade2win.co.uk/boards/showthread.php?t=9193 and look at post #1).
Nick

 

[TWI]

Rog,
Thanks for the post, some interesting results.
Do you have the total number of trades in each case?
I see that the Average trade makes a max of 13ticks. Am I right to assume this means that all trade PnL's added together divided by number of trades = what is shown in Average trade column?
If so then this will be tight once accounting for slippage, brokerage and b/o spread. I think it totally counts out the scenarios that only give an average 4-6 ticks.
Wonder how it would be if limit orders were used for entry which would assure a zero slippage on one side but may of course also mean that some good trades do not get filled and some bad ones too no doubt.

 

[RunTheNumbers]

Before anyone relies too heavily on 3rddawn's spreadsheet, I think there is an error in the code that calculates the stop loss/profit per trade.

RTN

 

[wellst2w]

RTN,
Are you writing about the problem referred to at http://www.trade2win.co.uk/boards/showthread.php?t=8333&page=3&pp=40 (see posts #81 to #85)? If so, I have the impression 3rddawn's spreadsheet results are in the right ball park. Or are you writing about another problem?
Nick

 

[rog1111]

Mr Chill

0.006 should read 0.0060, for some reason the end zeros are trimmed off. This equates to 60 pips. There may be other cases where only 3 dec places are given

rog1111

The

Nice work rog1111 ,interesting results
Re SL ,0.006 does this equate to 6 points or 60 ?

 

[RunTheNumbers]

RTN,
Are you writing about the problem referred to at http://www.trade2win.co.uk/boards/showthread.php?t=8333&page=3&pp=40 (see posts #81 to #85)? If so, I have the impression 3rddawn's spreadsheet results are in the right ball park. Or are you writing about another problem?
Nick

No I don't think so. The problem I was referring to was one I encountered relating to the stop loss limit. The trade was showing as a profit but should have been stopped out. I'm no VBA expert but I think the stoploss calculation on a short trade should check against the high point in each period, but actually refers to the low, and vice versa on a long trade.

RTN

 

[Mr Chill]

Cheers Rog ,once again many thanks for you work .

 

[rog1111]

twalker

I have updated the spreadsheet of results to include long winners & losers, short winners & losers & total trades :

Click HERE for results

You are correct in that the Ave Trade is the sum of all PnLs divided by total trades & I agree with you that the low average trades are probably unworkable. Since I am taking a fairly arbitrary time range on a chart, the high & low prices during the period could have been bid or ask, since they aren't necessarily peaks and troughs on the chart, but you will still lose the bid/ask spread once.

With a bit more tweaking, far better ave trades than 13 pips can be achieved, especially for $GBP, although I am aware of the limitations of using hourly data. Yes, I have also wondered about using Stop Limits to reduce slippage - apparently JonnyT uses them with a couple of pips leeway, and he has said that he hasn't had a non-fill from Globex so far this year. It's just a matter of time until it happens though I guess, probably in a fast market when you'd want to be in ! I guess the only approach is to try it with small size & compare with the theoretical.

rog1111

Rog,
Thanks for the post, some interesting results.
Do you have the total number of trades in each case?
I see that the Average trade makes a max of 13ticks. Am I right to assume this means that all trade PnL's added together divided by number of trades = what is shown in Average trade column?
If so then this will be tight once accounting for slippage, brokerage and b/o spread. I think it totally counts out the scenarios that only give an average 4-6 ticks.
Wonder how it would be if limit orders were used for entry which would assure a zero slippage on one side but may of course also mean that some good trades do not get filled and some bad ones too no doubt.

 

[jmreeve]

I have attached a Tradestation report for the spot on system back to 2001.

Results are for 1 contact include $17 per trade for slippage and comissions.

Data for testing was CME Globex traded Euro futures contract and was sourced from DiskTrading.

I have verified the futures data against spot currency data over the testing period and it seems to be accurate.

 

[garyp75]

I have attached a Tradestation report for the spot on system back to 2001.

Results are for 1 contact include $17 per trade for slippage and comissions.

Data for testing was CME Globex traded Euro futures contract and was sourced from DiskTrading.

I have verified the futures data against spot currency data over the testing period and it seems to be accurate.

 

[robber]

I have attached a Tradestation report for the spot on system back to 2001.

Results are for 1 contact include $17 per trade for slippage and comissions.

Data for testing was CME Globex traded Euro futures contract and was sourced from DiskTrading.

I have verified the futures data against spot currency data over the testing period and it seems to be accurate.

jim, thats a stunning bit of work. I dont know Tradestation but if its easy to configure it would be REALLY useful to see what returns you would get from running with the best of the variations identified by Rog1111s work;

i.e Take the breakout from 7-8 am, only trade it if it occurs by 10am and close ithe trade at 9pm (running with a fixed stop of 60 pts).

Neil

 

[Mr Chill]

Hi Rog1111

Just a quick question ,do you think this system would work with GBP/USD or is it specifically for the euro. Thank you .

 

[TWI]

jmreeve.
Since TS only allows order on close how did you handle the breakout order execution?

 

[jmreeve]

twalker-
Your statement is not correct.
Tradestation allows entry on a stop, a limit or at market on close.

For instance:
Buy ("LE") 1 contract next bar at 1.0123 stop;