How to configure an Algo-Finder on Vantage Signals
Return to the Vantage Signals user guide
In this video we cover configuring an automated strategy on Vantage Signals. Leverage our brute force algo-finder to find optimal algorithms for trading an asset and then tie them all together into a single strategy using our automated strategies. We offer an array of market profiling to further customize your strategy.
Strategies can decide on which of multiple algorithm signals to use based on current market conditions or recent performance and reliability of the underlying algorithms. We are taking a new, advanced and flexible approach to multi-algorithm, market aware, self-healing, automated strategies.
We can also enhance signal entry and exit by adding stop buy, trailing stop buy, stop loss and trailing stop loss to optimize your entry and reduce risk. There is no other platform that opens the door to technical analysis like Vantage Signals!
Below we cover, at a high level, how to configure and use the Algo-Finder to create and launch large batch jobs of algorithm configurations.
- When creating a new Algo-Finder, notice users can choose from the bull or bear tab. When searching for market pullbacks and short signals, users should create bear algo-finders. For long positions and market advances, users should create a bull algo-finder
- Click on the + to create a new Algo-Finder
When we first open an Algo finder, we are presented with the core configuration page. These are configurations common to all of the indicator combinations. Below is a screenshoot along with details about each of the fields.
- Name this template whatever you want, this is for your reference
- Choose one or more markets such as global, derivatives or a specific exchange
- Choose one of more assets to back test. Each market/asset will spawn a separate Algo-Finder Instance
- Tracking intervals: is for qualifying how well each signal performed from each algorithm configuration.
- Users can configure 0-3 Tracking Intervals and we will folow up this many minutes after each signal to assess how well the signal has performed. We track reliability and gains performance for each signal and provide rolled up performance reports for all combined signals for each algorithm combination tested. For example, a short term algorithm might report the signals performance 1, 2 and 3 hours after the signal where a long term algorithm may test 12 hours, 24 hours or 72 hours after each signal fires. The TI intervals are specified in minutes to 1 day would be 1440 minutes.
- The Bull or Bear Cooldown: This specifies that we should not send back to back signals, after we register a signal, we must cooldown this long before we can register another signal. This keeps passing market conditions from sending a signal every single interval being tested.
- The cooldown cannot be shorter than your longest TI period. We required completed assessment of the last entry prior to starting to track a new entry. Multiple values can be used with a comma delimited format such as 120, 360 which would provide a two hour and a 6 hour cooldown.
- Threshold states that all the indicators in the algorithm need to pass this many times in a row before we send a signal. This prevents us from sending signals on a market condition that only briefly exists. Using a threshold of 1 would state the first time all criteria match, immediately send an alert. This is defined in minutes. We can specify a range of thresholds.
- The step value says what value we should step through the range specified. For example, 1-10 with a step of 1 would check 1,2,3,4,5,6 and so on. Using values of 1-10 with a step value of 2 would check 1,3,5,7,9.
- There is no value in having a threshold shorter than your interval or else you will just test the same values since we couldnt have new values any faster than once per interval.
- The Intelligent intervals check box will automatically update the interval we use based on the shortest candle length in an algo. Depending on how many different candle lengths you are using across your indicators, this could save users up to 50% of the time it takes to process their algo-finder by using intelligent intervals.
- There is no need to test every 5 minutes if we are testing 1 hour candles that would provide the same value for at least 19 polling cycles using 5 minute intervals, so we auto use 1 hour intervals to remove redundant testing.
- The bottom section provides an option to configure a dependency signal. None of the algoritms being tested would be permitted to signal unless this dependency signal either is or isnt sending a flag or has sent one in the last x number of minutes. This is covered in its own article and explainer.
Once we have completed configuring the core tab, we can move on to the indicator tab.
We recommend reviewing the guidance for configuring various indicators in the indicators section of the Vantage Signals Userguide.
- We use RSA as an example here but its the same mechanics for all indicators.
- User can select multiple operators, all blue bubbles indicate how many options have been selected under each multi option selection.
- Users can enter stepped value where user enters a starting value, an ending value and then the step value says how big of steps we should take to get from the low value to the high value.
- A step of 1 would have us check every full value. In the example above -3,-2,-1,0,1,2,3. Had we used a step value of 2 it would have tested every 2 such as -3, -1, 1, 3. We also could have step partial values such as half percents where we would have checked -3,-2.5,-2,-1.5, etc.
- In conjunction to using a stepped value range, or instead of using a stepped value range, we can just enter a comma delimited list. If we configure both, we will test both ranges. Users shoudl take care not to duplicate values in these two sections.
- The example above would test -10,-5,-3,-2,-1,0,1,2,3,5,10 (NOTE: there are not negative RSI values but using this with percent change could look for RSI to fall a percent))
- Users can choose to test an absolute value or a percent change. With RSI we probably want to look for this value to go below 10, 20 or 30 or something like this, so we would look for an absolute value of 10, 20 or 30. Additionally with all indicators, we can change this from looking for an absolute value to looking for a percentage of change over a period of time. If percent change is selected, users can specify multiple comma delimited values for how far back (In minutes) we should compare the current value to look for this percent change.
- For example we could look for a 5% change (or any of the criteria values outlined above) in RSI over the last hour, or last 3 hours or last day and test this against all the possible combinations configured.
- Users can choose whatever metric they want to perform TA on, this doesnt have to be just price action, we can perform TA on volume, orderbooks, orderflow, market domination, social sentiment and more.
- Uses can choose one or more points in the OHLC(SA) candle to analyze
- Users can select one or more candle lengths to test from 5m, 15m, 1h, 4h & 1d candles.
- Periods defines the number of candles we should analyze with this indicator to produce a single value. For example, SMA with 24 periods using 1 hour candles would provide us a 24 hour simple moving average.
- The more periods the slower the indicator responds, the less periods, the more responsive the indicator is.
- Some metrics allow us to send a flag not when we cross the line but when we cross back over it
- Expiration will keep this signal alive for this many candle lengths before considering it having passed. This can be important when you have multiple indicators.
- With the example above, 3 candle lengths would be a 15 minute duration with 5 minute candle tests and a 3 hour window for 1 hour candle tests. For example, a user may want to set an expriation on a crossover and then give 30 minutes for us to also receive an alert on price and volume both moving within
After configuring the indicators and ranges desired for testing, all we need to do now is set one or more scopes of dates that we would like test. We have an updated interface already and will have a new article dedicated to Algo-Finder scoping like this one but for now this video covers configuring the back test dates and scopes in detail with the exception the new interface is cleaner and more modular.
Explainer video: Vantage Signals | Algo-Finder | Algo-Finder Scoping
Once you have successfully configured and run an Algo-Finder, you can review all the results in detail in the Visual Analyzer. See our explainer video and FAQ for the Visual Analyzer to learn more about reviewing the results of your algo-finders.