StockFetcher Forums · Filter Exchange · NEW SYSTEM WITH INTEGRATED 5% STOP LOSS<< 1 2 3 4 5 >>Post Follow-up
2 posts
msg #112269
Ignore ohbie1
3/15/2013 11:19:29 PM

OLDSMAR, Yes percentages. Here's the article. It seems logical to me. Of course there is a cost to this protection; you never get to 100% invested unless the stock continues down for a few days. I didn't have the patience so I risk a little more up front hence 33,33,33 is what I have been using. The good part is you can make up your own percentage to suit your patience/temprament/risk aversion. The bottom line is the 1-2-3-4 entry has a very high success rate.


In 2008 we introduced a concept that we labeled TPS, which stands for:

Originally presented as a stand-alone strategy, the core ideas behind TPS have turned into a methodology that can be applied to a variety of different mean reversion strategies. Since the overall concept has held up incredibly well over the intervening years, it’s worth a quick review.

With TPS, we’re trying to get all three components – time, price, and scale-in – working in our favor. At a high level, here’s how it works for a long mean-reversion strategy:
1.When a pullback occurs and the strategy signals an entry, buy a small initial position.
2.If the stock or ETF becomes more oversold (more on this in a minute), then increase your position size by scaling in, i.e. purchasing more shares.
3.Step 2 may be repeated multiple times if the price continues to pull back and the scale-in rules allow it.

Typically the criterion for scaling is simply a closing price that’s lower than the previous entry price. However, other rules may be used to determine when a stock has become more oversold. For example, we might enter a trade when ConnorsRSI is less than 20, and then scale in if ConnorsRSI drops lower. If the first scale-in occurs when ConnorsRSI is 17.8, then we would scale in a second time if ConnorsRSI closed lower than 17.8 at any time before we exited the trade.

Different strategies utilize different scale-in ratios. Common ratios include 1-1, 2-3-5, 1-2-3-4 and 1-2-3-4-5. The ratio is created by using each digit as the numerator, and the total of the digits as the denominator. For example, with 2-3-5 scaling, we can have an initial entry plus two additional scale-ins. The initial entry would be 2/10 (20%) of a full position, the first scale-in would be an additional 3/10 (30%) of a full position, and the second/final scale-in would be 5/10 (50%) of a full position.

The biggest advantage to this approach is that by scaling in at lower and lower prices, we are lowering our average entry price. This, in turn, increases our chances of a profitable exit.

A less obvious benefit of TPS is that it can allow us to enter trades using less stringent entry criteria. In many cases, this will increase both the number of trade signals generated and the average gain per trade.

Consider a very simple strategy with the following rules:
1.Buy a stock that has an RSI(2) value below X, where X = 10, 20, or 30.
2.Sell the stock when it closes above the 5-day moving average, MA(5)

Here are the results when the strategy is applied to a universe of liquid stocks and TPS is not used:

Here are the results when the strategy is applied to a universe of liquid stocks and TPS is not used:

Var #

# Trades

Avg % P/L

% Winners

RSI(2) Threshold




















As we would expect, as the entry criteria becomes more stringent (lower RSI(2) threshold), we generate fewer trade signals but a higher average gain per trade.

Now let’s see what happens when we use 2/3/5 or 1/2/3/4 scaling. In all cases, we scale in further when the price closes below the previous entry price.

Var # # Trades Avg%P/L % Winners RSI(2) Threshold Scaling
4 58463 1.38 80.02 10 2/3/5
5 81917 1.30 79.65 15 2/3/5
6 102258 1.24 79.61 20 2/3/5
7 58434 1.62 83.46 10 1/2/3/4
8 81917 1.54 83.21 15 1/2/3/4
9 102005 1.49 83.21 20 1/2/3/4

Notice that if we keep the RSI(2) threshold the same, then the number of trades remains stable, but the average gain per trade rises by a factor of 3 to 5. Alternatively, we can use a higher RSI(2) threshold to increase both the number of trades and the Average % P/L. For example, consider Variation 1, which uses no scaling and an RSI(2) threshold of 10. In back testing, this strategy variation generated 58,463 entry signals, and an average gain per trade of 0.42%. Variation 5 uses an RSI(2) threshold of 15 and 2/3/5 scaling, and generated 81,917 trade signals and an average gain of 1.30%. Using 1/2/3/4 scaling (Variation 8) increased the gain per trade to 1.54%.

As you can see, TPS is a powerful tool that can potentially increase the returns of many

152 posts
msg #112273
Ignore compound_gains
3/16/2013 12:41:26 PM

I don't think this is representative of forward trading, since the system was tested against the past 13 years and hit amazingly well. I am considering this a rough patch right now, which just happens to be right out of the gate...


I've been trading a variation of Kevin's original filter. December and January were very impressive; February was very modest and March so far has been a lost decent entries and breakeven or small losses on the ones I have entered. IMO there is something flaky about the market right now. My hunch is that when the overall market is strong these types of filters don't work so well for some reason? Probably we need some correction and volatility??

333 posts
msg #112295
Ignore jackmack
3/18/2013 8:06:57 AM

Good morning
I have been reviewing the charts of most of the successful trades from this filter and have noticed there appears to be a divergence in the RSI(2) from low that triggers to the previous low and a lower low RSI(2) reading for that low - so effectively a positive RSI(2) divergence WRT to a lower price.
It is not present in ALL but there is a greater percentage of those that display this.
I just thought this might help in the selection process you are looking to narrow down.
Thank you

4,548 posts
msg #112298
Ignore Kevin_in_GA
3/18/2013 9:08:34 AM

Not sure what you mean here - can you explain and maybe point to an example or two?

333 posts
msg #112301
Ignore jackmack
3/18/2013 9:43:08 AM

The RSI divergence appears when there is a previous low and an RSI(2) low BUT then when a stock is triggered IF there is a lower price low but NOT a lower RSI(2) lower that divergence "usually" leads to a better selection from the old eyeball filter.
An example is MTL - 4 March there was a price low around $5.20 and an RSI(2) around 3.3 but when your filter kicked this out on 13 March MTL was at a price low of $5.11 BUT RSI(2) was at @7.5 - a positive divergence.
It is not visible in every candidate but looking back it appears more often in those that have done very well since the filter kicked them out.
Hope this helps

333 posts
msg #112302
Ignore jackmack
3/18/2013 9:53:13 AM

MTL from Friday is another example

4,548 posts
msg #112304
Ignore Kevin_in_GA
3/18/2013 10:59:05 AM

The trick on this approach is that you do not know the RSI(2) until the end of the day, after you are in the trade.

However, you could add something like this:

set{entry, close*0.94}
"entry below low 10 day low"

This would insure that you would have a new price low but not sure how to select for an RSI divergence when you don't know what the rsi(2) is going to be ahead of time.

333 posts
msg #112305
Ignore jackmack
3/18/2013 11:10:35 AM

Ah yes - that would be a problem

4,548 posts
msg #112306
Ignore Kevin_in_GA
3/18/2013 2:01:36 PM

New trade triggered:

MM - 2663 shares at $7.51

104 posts
msg #112325
Ignore oldsmar52
3/19/2013 5:06:37 PM

If this is considered just a "rough patch" to go through, I'd hate to see what a "terrible period" would be? I would rank this system as one of the worst I've ever tried. When they hit the filter as a buy I got into ARNA, NUAN & WLT and am going to be stubborn enough to sit on them till I break even. System sounds good & may look good on paper, but it's a real stinker in real time.

StockFetcher Forums · Filter Exchange · NEW SYSTEM WITH INTEGRATED 5% STOP LOSS<< 1 2 3 4 5 >>Post Follow-up

*** Disclaimer *** does not endorse or suggest any of the securities which are returned in any of the searches or filters. They are provided purely for informational and research purposes. does not recommend particular securities., Vestyl Software, L.L.C. and involved content providers shall not be liable for any errors or delays in the content, or for any actions taken based on the content.

Copyright 2016 - Vestyl Software L.L.C.Terms of Service | License | Questions or comments? Contact Us
EOD Data sources: DDFPlus & CSI Data Quotes delayed during active market hours. Delay times are at least 15 mins for NASDAQ, 20 mins for NYSE and Amex. Delayed intraday data provided by DDFPlus