Market Filters Help
Concepts
A Filter is group of one or more technical indicators. A Filter is primarily used to scan the market to find stocks meeting it's criteria, for example looking for oversold stocks.
FilterScore is a numerical score assigned to a stock during analysis. For each indicator chosen all stocks in the market are scored and if more than one is chosen these scores are combined and normalized. FilterScore is always a value between 0 and 100, where a higher score means that stock fits the indicators(s) well.
FilterScore is meant to make technical analysis easier and more efficient. Traditionally analysts need to look at a graph and decide if a stock is about to break through it's Bollinger Band, as an example. Using the FilterScore a user can determine instantly how attractive a stock looks on the basis of Bollinger Bands without needing to process all the data yourself. The final score tells you how strong of a signal is being produced for any given indicator.
FilterRating is a concept combining the FilterScore of several filters into a single result. The rating is either "Overbought", "Oversold" or "Neutral". The filters are chosen based on their performance in back-testing, and the scores of the top few are averaged.
| Oversold | Neutral | Overbought |
|---|---|---|
 |  |  |
Stock Filters (and Scans)
These filters are run right away and the results will appear in the "Results" area below when finished. You can run multiple filters in a session which will then all appear in the "Recent Results" selector. They are listed with a timestamp so if you run the same filter again you know at what time. Using the "Recent Results" selector you can flip between recent filters you have run.
Stock Fitlers are a large table with the following fields:
| Rank | The rank of this symbol, starting at zero which is the highest. |
|---|---|
| Symbol | This symbol. Clicking on this will open a new window with more data at an external site. |
| Price | The latest known price of this stock. |
| Score | The Market Filters score that lead to it's ranking. |
| Open | The stocks open price today. |
| High | The stocks highest traded price so far today. |
| Low | The stocks lowest traded price so far today. |
| Prev Close | The previous days closing price. |
| Volume | The volume of trading for this stock so far today. |
| Avg Daily Volume | The average daily volume of this stock. |
| eps | An estimate of the Earnings Per Share of this stock. |
At the top of the results report are two extra number: Number Ranked and Average Score. Number ranked counts all stocks considered but not excluded by the indicators. Some indicators are primarily exclusion-based so this number may get small. Average Score is simply the average score of all ranked stocks. This number gives some extra context for evaluating the scores in the report.
Back-test Reports
Back-testing is used to test the relative effectiveness of a buy signal. The simulator begins at a specified date and begins to trade until the end date. It will hold one stock at a time and sell according to a customizable sell signal.
The sell signal is not very important here and may not implement the optimal trading strategy in every case. However, by using the same sell signal while comparing multiple buy signals the effectiveness of the buy signal can be measured.
It is recommended to run all tests over a period of at least a few months. Depending on the sell signal it may end up holding each stock for multiple days. If only a few trades are executed in a test that's not enough data to make a confident decision about a sell signal.
Intra-day historical data is available from September 2003 to the present.
The trading simulation used in back-testing is geared towards short-term holdings. The default sell-signal will be looking to sell a holding when any of the following 3 criteria are met:
- Profit exceeds 3%. 3% profit for a stock held for several days is a good target.
- Held for more than 20 days. This is to ensure no stagnant holdings, it could be making money elsewhere.
- Loss exceeds 15%. This is simply a safeguard to protect against any really bad situations. It's the weakest link in the sell-signal. In a purely automated system it's difficult to determine the best exit-point in these situations without knowing the cause of the decline.
Report details:
| Indicators | List of the indicators used as the buy-signal in this test |
|---|---|
| Starting Portfolio Value | The starting portfolio size specified to run the back-test |
| Final Portfolio Value | The ending value of the portfolio after executing the trades listed in the report |
| Start/End Date | When the simulated trading began and end |
| Transaction Count | How many buy and sell transactions were made |
| Wins | How many trades (a buy and sell) were net profitable |
| Wins after fees | How many trades (a buy and sell) were net profitable after taking the transaction fee |
| Losses | How many trades (a buy and sell) were not net profitable |
| Average Win % | For net profitable trades, the average percentage gain |
| Average Loss % | For net unprofitable trades, the average percentage loss |
| Transaction Fee | The transaction fee that was specified to use in the test |
| Transaction List | A list of all simulated transactions during the test |
Tuning Buy-signals
Using back-testing it's possible to measure and tune the performance of a buy-signal. If the win-rate is low (less than 50%) the set of indicators chosen is probably not effective enough on it's own. Try removing or adding indicators.
If the win-rate is good but the average loss % is much higher than the average win %, this is likely eating significantly into the profitability. This likely means the buy-signal is generally making good decisions but its bad decisions go really bad. These can sometimes be improved by restricting the space of stocks for trading. Try looking at non-penny stocks, just stocks with positive EPS, or small caps only.