Should I keep trading this system?

J

jedwards

These are the theoretical results from a system I've developed based on the DAX:
PHP: 
from                  to     profit/loss       win       full loss     small loss
2009-01-01      2009-01-31      -42.400000      6            5              8
2009-02-01      2009-02-28      -7.800000        9            6              5
2009-03-01      2009-03-31      32.600000       13          7              2
2009-04-01      2009-04-30      76.400000       13          2              5
2009-05-01      2009-05-31      41.900000        9           3              8
2009-06-01      2009-06-30      84.200000       14          1              7
2009-07-01      2009-07-31      95.500000       16          2              5
2009-08-01      2009-08-31      49.400000       10          2              8
2009-09-01      2009-09-30      26.600000       11          3              8
2009-10-01      2009-10-31      -7.700000        9            6              7
2009-11-01      2009-11-30      -1.800000        9            6             5
2009-12-01      2009-12-31      65.300000       13          3             4
2010-01-01      2010-01-31      -47.300000      5            6             8
2010-02-01      2010-02-15      26.400000       7            2             2
the profit/loss are points on the DAX, each point being 25 euros. The win/full loss/small loss are the number of each type of result I got per month. A small loss is after the trade reaches a certain threshold I move my stop up and if it hits I incur a small loss. I consider it "break even" but it ends up being about a 4 pt loss.

The risks I have with the theoretical results deviating from actual results are slippage, hitting my limits but not filling, and inaccuracy with my backtest data, but even with a 10% difference, the results are still pretty good (assuming a $10k funded account, it returns over $10k since Jan 09).

It's purely algorithmically driven, so I don't sit in front of the computer, I wrote a program that trades for me. It trades once a day only, so I don't risk overtrading.

After working on this system for a few months and backtesting it, I sim-traded it in December. As you can see, the results were outstanding, so I had a lot of confidence in it. In January, I traded it for real, 2 contracts, consistently for the month.

As could be expected given my luck, I got my balls handed to me. It was a terrible loss, and it shook my confidence in the system, so I stopped trading it at the end of January. As could be expected given my luck, just as I stopped, it looks like the system is recovering for the month of February.

My question is, should I keep trading this system? My confidence in it is completely shaken, even though it looks like it's working again. Mentally, I can't take any more losses, the relentless losses throughout January have significantly damaged me. The drawdown I took just in trading the DAX was significant, but I traded across several other futures contracts as well, and had similar results, leading to a 25% drawdown in my account in the last 2 weeks of January. While the others aren't doing as well in February, the DAX seems to be doing okay now.

Mentally and psychologically, I can't stomach another drawdown. I'm worried that if I start trading this again, it will start to tank. I know it's irrational to think that the markets are some how conspiring against me, but financially I can't afford any further losses.

Any suggestions or guidance would be much appreciated!
 
Not enough data there to draw any statistical conclusions. Try producing results for at least 3 years. The only conclusion I drew is that you should not trade during the month of January:)
 
Your pnl just looks like buy and hold perhaps levered more or less during certain periods. Do you buy as often as sell? If not, backtest over down periods as well.


Quote from jedwards:

These are the theoretical results from a system I've developed based on the DAX:
PHP: 
from                  to     profit/loss       win       full loss     small loss
2009-01-01      2009-01-31      -42.400000      6            5              8
2009-02-01      2009-02-28      -7.800000        9            6              5
2009-03-01      2009-03-31      32.600000       13          7              2
2009-04-01      2009-04-30      76.400000       13          2              5
2009-05-01      2009-05-31      41.900000        9           3              8
2009-06-01      2009-06-30      84.200000       14          1              7
2009-07-01      2009-07-31      95.500000       16          2              5
2009-08-01      2009-08-31      49.400000       10          2              8
2009-09-01      2009-09-30      26.600000       11          3              8
2009-10-01      2009-10-31      -7.700000        9            6              7
2009-11-01      2009-11-30      -1.800000        9            6             5
2009-12-01      2009-12-31      65.300000       13          3             4
2010-01-01      2010-01-31      -47.300000      5            6             8
2010-02-01      2010-02-15      26.400000       7            2             2
the profit/loss are points on the DAX, each point being 25 euros. The win/full loss/small loss are the number of each type of result I got per month. A small loss is after the trade reaches a certain threshold I move my stop up and if it hits I incur a small loss. I consider it "break even" but it ends up being about a 4 pt loss.

The risks I have with the theoretical results deviating from actual results are slippage, hitting my limits but not filling, and inaccuracy with my backtest data, but even with a 10% difference, the results are still pretty good (assuming a $10k funded account, it returns over $10k since Jan 09).

It's purely algorithmically driven, so I don't sit in front of the computer, I wrote a program that trades for me. It trades once a day only, so I don't risk overtrading.

After working on this system for a few months and backtesting it, I sim-traded it in December. As you can see, the results were outstanding, so I had a lot of confidence in it. In January, I traded it for real, 2 contracts, consistently for the month.

As could be expected given my luck, I got my balls handed to me. It was a terrible loss, and it shook my confidence in the system, so I stopped trading it at the end of January. As could be expected given my luck, just as I stopped, it looks like the system is recovering for the month of February.

My question is, should I keep trading this system? My confidence in it is completely shaken, even though it looks like it's working again. Mentally, I can't take any more losses, the relentless losses throughout January have significantly damaged me. The drawdown I took just in trading the DAX was significant, but I traded across several other futures contracts as well, and had similar results, leading to a 25% drawdown in my account in the last 2 weeks of January. While the others aren't doing as well in February, the DAX seems to be doing okay now.

Mentally and psychologically, I can't stomach another drawdown. I'm worried that if I start trading this again, it will start to tank. I know it's irrational to think that the markets are some how conspiring against me, but financially I can't afford any further losses.

Any suggestions or guidance would be much appreciated!
More...
 
Quote from jedwards:

...Mentally and psychologically, I can't stomach another drawdown. I'm worried that if I start trading this again, it will start to tank. I know it's irrational to think that the markets are some how conspiring against me, but financially I can't afford any further losses.

Any suggestions or guidance would be much appreciated!
More...

Yeah, I think it's odd that you only backtested back to January 2009 considering you want to use it for automated trading because maybe your system still performed well overall even with a few drawdowns had you tested it further back more years.

Also, you show an excellent example that befalls most discretionary traders...emotions or lack of discipline via the fact that you stop trading it in January as soon as you hit your first real money trading drawdown. This is important information for automation traders to understand that it still takes a human person to apply the system just like any discretionary trader. However, you said something that really caught my eye. If you can't handle a loss as in that the money you lose will cause you financial hardship...

Don't trade and go back to the drawing table to test your method going back more years because there just may be enough information with more testing to give you confidence that the drawdowns are part of the game while remaining positive by end of year (each year).

Mark
(formerly known as NihabaAshi)
 
Two things are evident....

1) It seems that you are underfunded and have limited resources to trade the DAX...by your stated emotional concerns.

2) While your statistics are limited, it is patently obvious that you should have been trading the system in reverse...those small losses would have been profits a long with the major losses...vs stops being your current "winners"

Go back over your data and if it is less than 60 -65% successful consider reverse trading the system with proper money management controls....and in the meantime...rebuild your funding to the point you will have adequate funds to trade.

Can you apply this system to NQ and or ES...maybe less pressure for you to endure...


Good luck

NiN
 
Just out of curiosity, what is a "win" and a "full" loss? You said a "small" loss ends up being about 4 points. What is the ratio of a win to a full loss? Are your profit targets fixed, or via a trailing stop? Your winning percentage seems pretty decent, so unless you are taking quick profits, it seems like your system should be doing quite a bit better than it is.
Also, for the small loss scenario, what about moving your stop for half to BE and the other half to -4? This way, if you are hit for a small loss, you actually only lose 2 points on average rather than 4.
 
Thanks for everyone's answers!

@intradaybill and wrttrader: I think you guys are right. Now that I step back a bit, ~1 year's worth of backtest isn't enough, because it doesn't give me enough confidence to see if the system can bounce back after a significant loss. I will take your advice and try to acquire more data and backtest at least 3 years. I think that is the significant factor in the equation is that I don't have enough of a long-term view on things. I thought a year was enough, but it's not. I have enough to know that it was profitable, but I haven't seen it really do well, have a couple of crappy months, and then bounce back. Thanks!

wrttrader, with respect to it being a financial hardship, I guess I was being a bit dramatic but it did wipe out 25% of my savings, which is significant for me. I started this system not with the goal of hitting 100% returns, but with the hopes of having 1 more win than loss per month, so I was targeting a 2% profit per month or ~20% profit per year. Based on what I had seen in my backtest, it seemed reasonable to expect this, but I think you're right in that I need to have longer view in terms of backtest.

@noddyboy, the system I use goes either long or short, and it's a day trading scheme so I hold from anywhere from a few seconds to several hours. I think the ratio of buying to shorting was 60/40 but I would have to double check.

@Now is Now: Thanks for the advice. I think however, that the numbers show mostly profits for the entire year, meaning that the wins outnumber the full losses + small losses, for most of the year. Reversing it would lose money, from what I can see. In terms of me being underfunded, that could definitely be the case, but especially now given my drawdown :( I guess the size of my funds might be a lot smaller than most of the professional traders that frequent here often, so when trading 1 or 2 DAX contracts, a string of 8 losses would have a severe impact in my funds.

@The Goonior: Sorry, I mistyped, I meant 4 ticks which is about 2 pts. My average full win/loss is about 8-10 pts on average and I keep my target to stop loss at 1:1. When I have a certain amount of profit, I move my stop to -2 pts from entry, and then continue to wait for either the target or stop to get hit. It certainly seems like there is more I can do to play around with the stop, it's rather coarse right now, and if I can change those small losses to either break evens or small wins, then it would make a big impact. I find however that it tends to impact the system overall so it will take some more time to devise something better.
 
Quote from Now is Now:

2) While your statistics are limited, it is patently obvious that you should have been trading the system in reverse...those small losses would have been profits a long with the major losses...vs stops being your current "winners"

Go back over your data and if it is less than 60 -65% successful consider reverse trading the system with proper money management controls....and in the meantime...rebuild your funding to the point you will have adequate funds to trade.
[/B]
More...

When I *first* started developing algorithmic trading strategies, I thought just like you did. Why, if you were losing on 70% of your trades, couldn't you just reverse the system?

In fact, how many traders have said, "Man, I should just do what I normally do, except OPPOSITE?"

It just doesn't work that way.

Most algorithmic systems are not symmetric. You do not buy at x, then pray it goes to X+2 or X-2.

Furthermore, there's an order-of-operations issue - that is, which is the market likely to hit first - your target or your stop? Just because you've reversed, doesn't mean that the reverse system won't produce the same - or worse- results.

Trust me.

If finding the holy grail of trading systems was as simple as reversing systems with 70% losers, I have 32 trading systems I'd like to sell to the community here.

Regardless, good luck to the OP.
 
I understand your point, however I was referring to 60 -65 winning trades not 70% as that would be too difficult to over come .

The point I was trying to convey, apologies if it was not clear, that where the system was seeking a 'small profit' level that could well be the entry point to go in the opposite way..the stop will be the ' higher profit' level.

If the system was really robust the OP would not have brought up the issue in the first place...

Also, I did preface my comments relating to the limited amount of data...ideally it would have been interesting if we knew the max winning and losing sequences and win/loss patterns...also, of course, the range of profits vs losses..were they equal?

As I said we were "flying "with limited information, but the OP did concede the most critical factor to his concerns....limited funding...

NiN.
 
You must log in or register to reply here.
Back
Top