Chabah on Automated Trading

Here is an attachment for just January. As you can see, all the profits are long - this shows that the system does a decent job of following the market in an uptrend. What remains to be seen is if the trend system can turn a profit in a down or churning market, or on particular stocks.

For some reason this summary doesn't show the real Open P/L - it is around $290 this morning.

Enjoy,

Chabah

 

January ended better than I could have possibly hoped. Trading 12-14 stocks (I added and removed some as I adjusted during the month) with an average risk per trade of $50, I had 62 trades with an average profit per trade of $37.85, or almost .75R. So in 62 "bets" with an expected average loss of $50, we won 42% for a total of about 27 bets or $2350.

The actual average loss was $55 thanks to a $525 loss on Heely's that really juiced the number.

So, that is pretty much what I expect as a favorable result. However, this far understates the results because of the open O/L, which is at $2240 at the market close. The system remains short NTRI, which has been a huge winner with open P/L of almost $2k. An open AAPL short has about $300 of profit. These positions will, by definition, give back some gains before closing.

So, including open P/L, the result for the month is about $4500. Given the average risk of $50, that's a 90R month - unbelievable.

Honestly I think this will likely be the highest R return month of the year and perhaps for several years. There is clearly a large luck component, which will be impossible to duplicate reliably. That said, I think the system may be able to grind out an average 1.5 or 2 R per day which would be just fine with me.

With my account above $36k, $50 as my R represents about .14% risk per trade. On most of the stocks (not Heely's or a couple other shaky issues) I am increasing the average loss to $75, or .2%. Eventually I would like to get to .33% and then .5% once I am really confident in the system.

Note that unlike many systems' use of R for risk, my system is using R as the average loss reported through backtesting for 1 year. Except for the bar when I enter a trade, there is no stop loss order limiting my loss to 1R. As in the case of Heely's, I took a 10.5R loss on a market gap - that can happen. It is also why I simply can't use higher risks such as 1% per trade - the actual account risk would be way too large.

So wish me luck for February. I will post some of the Tradestation graphs and such after NTRI closes and they show the full picture. I would be happy to answer questions regarding the system - just post them in the forum and I will do my best to provide an informative response.

Thanks,

Chabah

 

Good work, C.

With an automated system which is consistent, methodical and and adapts to the conditions in the marketplace as they occur.

That's what good trading really is, you know?

How scalable is your system? and at what Level do you beging culling profits?

Good trading,

JJ

 

Thanks, JJ.

The system is scalable in the sense that it does not require much work, so I could pretty easily monitor more stocks. The number of stocks that can be traded is directly proportional to the risk per trade, so if you had say 50 stocks you would have to limit risk to say .1% of equity in order to avoid capping out your margin. I am sensitive to this as I have been so far unable to get TradeStation to limit my margin to 100% of account size (it gives 300% intraday margin, but most trades are multi-day).

Personally I would rather trade the "best" 12 or 16 stocks with a larger risk per trade (.5%) rather than risk very little trading a bunch of stocks (say .1%). My buddy is using a similar but manual system on almost 100 stocks, but we don't have numbers on the performance yet largely because he was out of the market on days he couldn't attend to his system. One big advantage of the automated system is that there is hardly any reason to be out of the market (power failure, etc).

The other side of being scalable is that at some point the trades could move the markets. However, I think I am very, very far from any such situation. That would probably require an account in the multiple millions, and even then the stocks I am following have HUGE market caps that are unlikely to notice me - AAPL, CME, NYX, PMTI, NTRI, BIDU, etc. Pretty much only institutions move those stocks.

If the system continues to earn consistent profits measured in R, then it makes sense to try to get R to be as large as possible - i.e., to grow my account. Therefore, I am not intending to cull profits until the account is at $200k or so. If I can cap it at 100k and pull out $2k per month, it would be better to cap it at 200k and pull out $4k, right?

At $200K I will probably need to diversity my investments a bit, but the account should still be grown to $1 mil or more. But we are talking about years of growth and we only have 1 successful month, so I am not exactly planning in detail. I could easily be out of business in April.

So for the time being as the account grows I will simply scale up my position sizes such that 1 R (average loss) stays in the .2 to .5 range as discussed 2 posts ago - that should limit the risk of going broke while putting most of my account to work most of the time.

Thanks,

Chabah

 

Here is an up to date chart of AAPL. This trade is still open to the short side, so that profit is not shown - it is currently around $11 per share profit. You can see how the trade reversed from long to short when the uptrend broke on 1/17. Apple's trends require 5 bars to qualify as a channel.

You may also note the very wide channel that I use for AAPL - it is approximately 1.6 times the 20 day volatility. This allows a full 6 days of sideways action - some might prefer the trade to end here, but that is not what the backtesting recommends for AAPL.

Note that because of the wide channel, the position size is quite small. The position size is directly proportional to the width of the channel. The difference is evident when compared to NTRI (next post).

One interesting feature of the system is displayed on 1/30, where the channel re-adjusts instead of breaking (and reversing). This occurs when one channel is broken but an valid channel still exists in the same direction (long or short). This readjustment will allow more time for sideways action, and so far has yielded about another dollar per share gain.

 

Attached is a chart of NTRI - again this trade is still open so the long blue profit line is not shown. The interesting thing about this chart is how the trend - which is relatively flat - readjusts to the stock after the big gapping bars. A version of the system that increases the slope of the trend would have been knocked out on 1/26 and 1/30. By maintaining the initial modest downtrend throughout the gaps, the stock is held much longer.

Gaps always present big risks, but they are also oppportunities if you get them right. It seems that for January at least the trends indicated the gaps at least more than 50% of the time, so the gamble paid off. We'll see how the luck factor plays out over time.

Thanks,

Chabah

 

The streak continues here in February to the tune of about 17 risk units or about $1300. Again my risk units are set at approximately .2% of equity or $75, and we'll be looking at .33 for March if February looks solid (no major technical problems). I have already noticed a little heart pounding with the increased position sizes, so changing bet sizes does require some mental adjustment.

Today's burst was courtesy of PMTI, which bought in around 52.70 on the open (based on yesterday's trend) before Palomar exploded today. Very nice when a plan comes together (shout out to Dan Fitzpatrick on The Street!). It will be interesting to see what happens with earnings tomorrow, and if system is in the stock at the end of the day. A bad gap could easily vaporize a few gold nuggets.

I await the steely slap of variance, and it is inevitable that there will be some big losing days, probably weeks, and perhaps even months. But I'll enjoy this while it lasts.

Chabah

 

That didn't take long! PMTI = OUCH.

Oh well, dems da breaks.

 

Have you considered exiting positions before planned announcements or do you think over time these announcements will average out?

 

I think the trading action before an announcement will indicate the direction of the gap around 50% of the time or more, so I am not looking to avoid announcements or events.

In this case my system actually wanted to sell and reverse based on the action at the end of the day, but the orders were entered at 4 pm and didn't execute. This is a system flaw and I will adjust the system to create end of day orders at 3:59 instead of 4 pm. In this case such a change would have saved me almost 20R, so it's a pretty important change.

Thanks,

Chabah