Delphi Universal is a completely mechanical swing and day trading system which was released September 27, 2005. In May, 2007, Delphi II was introduced. The system was simplified, and walk-forward optimization (see the discussion below) was used to generate the out-of-sample performance results, as well as the current trading parameters.
Most systems are developed with one market and timeframe in mind, making over-curvefitting much more likely. Some systems are able to trade a number of markets within a market segment and/or a specific timeframe. What makes original Delphi unique is that with exactly the same rules, it successfully trades a wide variety of markets and timeframes, from intraday domestic and global index data to daily data of every sector of the domestic and global commodity markets to Forex. Few systems available to the public can make this claim.
Delphi original is a robust trend-following system that employs a dynamic channel breakout entry technique. A proprietary filter reduces the likelihood of trades during congestion periods. Entries may occur on a price breakout or retracement from the main trend. In order to avoid over-curvefitting to one market, the system was developed using intraday data of the 5 main US indices and several non-US indices. End-of-day data was used to test a basket of 42 non-index commodities in every sector. Although Delphi does not net a profit in all markets, the fact that it is successful on a wide range attests to its robustness.
Delphi II is the result of applying a proprietary walk-forward optimization routine to original Delphi. The result is a simplified system with a hypothetical performance record that is virtually equivalent to a real-time record. At present, it is most recommended for swing trading EM (the emini Midcap), day trading ER (the emini Russell), and swing and day trading the DAX, since these markets are the best trenders of the major indices, and they have good volatility and dollar-per-point value.
All entries and exits are stops or limits. All exits are based upon volatility, as determined by ATR (average true range). There is one profit lock level, where the stop is moved up to protect a profit, and a profit objective limit. The objective is reached about 20% of the time, avoiding some of the give-back that trend-following systems inevitably experience. The system reverses about 50% of the time. Delphi trades an average of 7 trades/month per index and is in the market about 55-60% of the time.
Lease, Purchase, or Lease-to-Own
Delphi Universal can be leased for $50/month/contract, purchased for $2995 (open source), or leased-to-own over 5 months and 6 payments of $575 each. Any leases are for Delphi II, but a purchase includes both original Delphi and Delphi II. A purchase also includes the following: for a period of five years from your purchase date, provided that you have sent TradingVisions your updated e-mail address, you will receive by January 5th of each year the updated parameter values for the recommended markets, or, at TradingVisions’ choice, you will receive the walk-forward optimization instructions that will allow you to determine the best parameter sets yourself.
Performance Information
To view actual & hypothetical performance reports, click here.
Why trade both Delphi & AXIOM Index? Click here for the answer.
Walk-forward Optimization:
The hypothetical results below were derived from walk-forward optimization (WFO), and they are therefore as close as we can get to actual performance data. Optimal parameters were derived from past data ("in-sample," "study," or "backtest" data), then applied to the next year of data (this "future" period is "out-of-sample" or "application" data). The process then moves forward, with what was the "future" year becoming part of the "past" (i.e. the application year becomes part of the study period), and the next year being the new "future" or application year. For example, say we have data for 2000 to present. We could use 2000-2002 as our study period, run our backtests, and find the parameters that give the best results. We then test these parameters on the data for 2003, and these become the first year of our performance record. These results from 2003 are categorically more important than 2000-2002, because they are results we could have actually achieved (assuming we are back in the year 2003), whereas the results from the backtest period (2000-2002) could not have been achieved, since they were in the past. We next "walk forward" to include the 2003 data in our new study period (we can either add it to the prior period, so we now have 2000-2003, or we can drop the first year and study 2001-2003). We find the best parameters for the new study period and apply it to our new "future" year, which is now 2004. The performance from 2004 is added to the first results from 2003, expanding our "real time" performance history. We then continue to walk forward, adding the next year to our study period, and the next "future" year to our performance record. While the great majority of performance reports are idealized and unrealistic because the backtest or study period is the only data that was used to generate trades, a walk-forward hypothetical performance history is as close as we can get to actual performance numbers because it is composed of results that are from the parameters derived from past data, but applied to "future" data.
Because of walk-forward optimization, Delphi II has the second advantage of being adaptive. Each year by January 5th TradingVisions will use the new parameters that incorporate testing that includes the previous year’s data. In this way, Delphi II will be able to make adjustments to market changes.
For further information on walk-forward analysis, click here.
The improvement WFO has made in Delphi is most dramatic in a market-change year like 2006. For the e-mini Midcap, original Delphi made $700 in 2006 and is down $3,000 this year, through the end of June (this is using the TradeStation report, with $50 slippage/commission), with a max drawdown of $10,000. Delphi II would have made $9,080 in 2006 and $1,465 this year, with a max drawdown of $6,200. Keep in mind that these results for Delphi II are not idealized, backtest-optimized results, but are walk-forward, using the rules and parameter values determined from testing prior years, so this is a fair comparison with the original Delphi results.
The equity curves below are entirely based upon out-of-sample data & are therefore a close equivalent to real-time results..
