My back-testing is too good to be true. What am I missing?

T

TradeSparrow

Please see the attachment for my spreadsheet.


I'm back-testing a trading system and, as you can see from the spreadsheet, the results are spectacular. It's hard to believe I am doing this right. Can you guys help me find issues with how I am trading? I'd be really grateful!


I'm back-testing my strategy by clicking through historical Wheat charts one day at a time. The data is from Think or Swim, and I'm starting with Wheat in 2003.


I posted this question on Reddit, but there seem to be no AG traders there. Many brought up some questions, and I'll summarize my answers so we can get some basics out of the way:


-- How do I decide when to enter and exit?
I look for trends and large reversals on daily charts.
These are the indicators/patterns I use:

1. EMA and MACD-Histogram to find trends.
2. Heikin Ashi Candles to find trends.
3. MACD-Histogram and Price Divergence to find reversals.
4. Engulfing Candles to find reversals.
5. Compare price movement to Elder's non-EMA'ed Force Index to determine conviction.
6. Historical highs and lows to determine whether the price may be at its very bottom or very top.


-- How do I actually enter a trade?
I use a trailing stop limit set $2 from the price (Bid/Ask depending). This helps me get in near an intra-day high or low if I'm right.


-- How many contracts do I buy?
For every $5000 of capital, I buy one contract.


-- How much do I risk per trade?
I place my stop $2 away from my entrance price, which is a 2% risk, not including commissions. Including commissions, the risk ranges from just under 2% to 2.5%.


-- Is $2 enough wiggle room?
Yes, it seems to be enough.


-- Do you use limit or market orders?
I use limit orders, even for my stop loss. However, behind my stop limit order, I put an emergency stop market order just in case the market explodes and my limit order isn't hit.


-- What's my plan for taking money out for personal use?
I don't have one set in stone. I'm more concerned with making profits first.


-- Have you forward-tested the system?
Not yet. Forward-testing begins tomorrow when the markets open.


-- Do I tinker with indicator parameters a lot?
Not once since the beginning of the test.


-- Do you know when your system doesn't perform well?
The system has trouble during ranges. I switch to shorting at the top and buying at the bottom when a range occurs.


-- If the market jumps I lose 15%, do I leave or do I ride it out?
I leave and re-think what's happening.


Other thoughts:
- How liquid are the grains markets? Can I easily trade 10-20 contracts?
 
Would help if you described what you think is 'too good to be true'. Its fairly easy to get an extremely smooth equity curve, but if you are not projecting over 1-2 cents profit per share traded, chances are you are over estimating on your backtest due to simulation shortcomings.
 
Out of the top of the head you might:
1) Forget about bid/offer spread. If your buy-limit order is at the current bid price, there is no guarantee the order will get executed even of there is substantial trading valume at the bid.
2) Use long bars so that buy and sell orders are executed within the same bar. There is no way to know if open-position order gets executed before teh close-position order without looking at a finer time frame.
3) You may be ignoring commissions.
 
In terms of slippage considerations, if you'd like to see a live screen-shot of an order book DOM let me know either here or on PM and I will copy and paste them for you.
 
I'm not sure what you mean.

Quote from walterjennings:

But if you are not projecting over 1-2 cents profit per share traded, chances are you are over estimating on your backtest due to simulation shortcomings.
More...


Unless I am misunderstanding you, the "not" is a typo, right?


What I mean by "too good to be true" is that the first year, I have over 800% profits recorded. The second year, over 2800%. The third year, over 1500%. If that's not too good to be true for you...would you be my mentor? :D

Each year, I reset the starting capital to $10,000.


Even if I divide my results by 10, the profits are still ridonkulous. :confused:
 
If anyone is interested in the Reddit post, you can find it here.


Thanks LeeD!

1) This point is really helpful. I understand what the Bid/Ask prices represent, but I don't know if I should be using them to decide where I enter and exit.

What price should I have my orders execute at? Mark? I can also link my stop and limit prices to Last. Which I should be using.


2) I'm not sure I understand.

I use Think or Swim, which allows me to set triggers. For example, if my open-position order is executed, a close-position order will be immediately created.

I forward-tested the entry strategy over several days with success. Today, I back-tested the entry strategy over the past 20 days using 5-minute charts. Again, it worked very well.


3) I'm including commissions. :cool:


Quote from bone:

In terms of slippage considerations, if you'd like to see a live screen-shot of an order book DOM let me know either here or on PM and I will copy and paste them for you.
More...

Bone, that would be fantastic.

However, should slippage be a big concern? I use mostly limit orders to enter and exit. For exiting, I keep a back-up stop behind the closing limit order. Hopefully, if price skips over my limit order, the stop will prevent too much loss.
 
You must log in or register to reply here.
Back
Top