Financial Market Modeling Projects

Raymond Raud, May 2009

This is a selection of short term proprietary trading algorithms that are worth attention and I have had time to write up. Brief summary of reasoning behind each algorithm, market conditions it works well and market conditions it does not are below. Few others are in works, some more complicated, but promising ones already for years.

Most traders are inclined to trade breakouts rather than consolidations. The industry literature cliché’s “trend is your friend” etc. promote that idea although they occur relatively less frequently then consolidation trading opportunities. Understanding the popularity of break-outs first algorithms are for these strategies.

Time Breakout

As it turns out each market has a specific time period when the traders establish their short term direction bias. The direction bias is expressed as a price change during that period. Same directional bias frequently enough dominates the price move for following trade period creating a profit opportunity. Some think the bias period is always at market open, but this is incorrect. The bias period and the length of profit period differ by weekdays for the same market. For some weekdays the bias period is aligned with economic news usually reported at specific times, for other days it is loosely correlated with market open. It should be understood that for globally traded markets like energy and currencies there are at least three “opens” for each calendar day. Difference in trader’s time zones is an additional factor influencing the bias period and blurring the start-end of the period. Also, for some markets and some weekdays the bias is not strong enough to warrant risking a trade.

Knowing the bias period and trading period for each weekday and each market makes this strategy very easy to implement. Following is what one needs to do:

1. Establish the high and low prices during the bias period – the entry range

2. Enter in direction of the break-out from this range

3. Close the trade at the exit time

   At most there are three meaningful trading periods: Asian, European and American. Last two overlap. Except for Crude Oil only European and American trading periods are meaningful with this strategy. Hence, depending on the market and day of the week one may have to implement one or two trades. Of course, a trader can chose to implement only the most convenient trading period. For a simple strategy it is surprisingly consistent and profitable except for some markets. Of course, it can be unpredictable at turbulent times. Some markets are much more consistent than others, but substantial draw downs should be expected on all of them.

   Below are simulated equity curve examples for some markets (cost of trade not included). All parameters (bias period start, end, weekday schedules and trade exit times were derived from the period immediately preceding depicted on these graphs. In other words, what you see is not a result of retroactive curve fitting or optimization.

   
   Worth noting the turbulent period on this market was from mid Sep 2008 until early Mar 2009. Whoever remembers, it was not complicated decision to stop trading for the period of extreme anxiety and uncertainty.

Another conclusion from this analysis. The traders on the GBP market must be generally convinced in correct direction (stubborn). They either rarely change their mind during the trading period independent what happens … or majority of the time nothing really influences this market.

An example of a market you may not wish to trade with this strategy is here. Again, all parameters were derived from period immediately preceding depicted on the graph.

Trading Noise

Noise is generally considered the enemy of a trader. Due to largely statistical nature of market price discovery it exists on any freely traded market. Its existence and nuisance it creates for traders has motivated development of a large number of technical trading indicators. A moving average is probably the oldest and most widely known among them. The objective of all technical indicators is to suppress the noise in price moves revealing the underlying price move direction. As opposed to fighting noise this strategy uses it for profit. The strategy is fully mechanical and very simple. Human intervention is only required to switch it off when the market conditions turn adverse.

The strategy works well on slow moving very deep markets with ample noise in the price discovery. It can lose a lot when the market moves strongly in any direction. Treasury markets and S&P 500 E-mini during periods of approximate equilibrium are good for this strategy. Profitability disappears once the volatility increases, anxiety settles in and forces rapid irrational market moves. The strategy generates large number of trades with small profit (few ticks each). Essentially, low commissions, pricey tick, automatic order entry and rapid execution are necessary for profitability.

Trade list for S7P E-Mini Se 2009 contract June 30, 2009 is a good example of benefits and pitfalls of this strategy. Trades are simulated from 1-min bars, profit/loss is in points (each point is $50.00). Note the 10:00 – 10:36 trade with substantial loss after Consumer Sentiment report disappointed traders. Essentially, if one would use this strategy for real money one would shut it off for the time of critical economic reports and news. Without the 12 point loss the day would have been pretty good.

Gold Digger

One of the break-out strategy deficiencies is loss of profit from the move through the range. This loss can be substantially larger than profit from break-out when the range is wide. Early detection of price move turn-around is paramount for success in range trading. Searching for a solution for both I stumbled on a market’s own early warning signal for a future move direction. Further research led to the discovery of another early warning for likely end of a market move. My scan of past industry literature has so far not produced any indication of these phenomena being described by anyone. With my electrical engineer background I’ll call the entry side signal “voltage surge” or vS for short. The end signal indicating the end of the move potential will be “voltage resolution” or vR for short.

Both vS and vR are present, readily available and obvious from daily and intraday chart data. For visualization, the data needs to be presented differently than it is usually done in the industry. This is specifically true for intraday chart data. The transactions on the market are regularly aggregated by fixed time intervals (1 min, 5 min, … 60 min). Yet, the beginning and the end of this aggregation period differs from server to server and desktop to desktop. They are arbitrary and have actually no relevance to the trader’s decision making.

With a number of refinements Gold Digger strategy is built on vS and vR. Since these signals are derived from the current market behavior rather than smoothed and averaged past, theoretically it should perform well on any market conditions and on any market. Yet, this project is still work in progress. Even though indications so far are positive it is too early to tell whether my theory holds.

For a snapshot of what this strategy is capable of here is a list of trades from June 24, 2009 from 12:50 until 16:40. Note that FOMC came out with their long waited statement during this period (14:15 as usual). This is an event the usual mechanical strategies prefer not to trade over.

Entry Time	Entry Price	Long Short	Exit Time	Exit Price	trade P/L	Cumu lative
12:56:34	932.1	1	1:28:14	933.6	$150	$150
1:37:35	934.7	-1	2:19:55	934.5	$20	$170
2:22:45	936	-1	2:41:26	930.1	$590	$760
2:55:58	930.6	-1	3:09:43	928.5	$210	$970
3:14:28	929.5	1	3:30:32	930.9	$140	$1,110
3:35:56	930.4	-1	3:45:32	930.7	($30)	$1,080
3:46:18	930.5	1	4:26:00	932.3	$180	$1,260

Back to home page

© 2009 Raymond Raud. All Rights Reserved.
Last Modified: May 2009