- Algorithmic trades can boost your success rate, but they don't work on hunches
- Your trade programme will need continuous optimisation
- Look to decrease risk as your trade size grows
- Consider a basic algorithmic trade based on how EURUSD moves vs GBPUSD
- Other 3-way currency pairings can also be established with this simple, robust setup
- Set stops as a matter of default; eliminate lag as much as possible
- Beware the potentially destructive spread between futures and cash-market prices
- Ignore anyone who says the Monte Carlo technique works
May "the force" be with your algorithm. Image: iStock
Let's talk about "algos" — algorithmic trades as a way to boost your chances of consistent success and as a means of defending your cash pot in case the market goes against you.
First I need to assume we know our risk structure. Just how much are we really prepared to lose? What percent of our total pot of trading cash are we willing to use per trade? And from that we should know exactly how many trades we can make before getting wiped out. Once you know this figure — and only then — you can start considering building your first algorithmic trade because any trade programme you devise must have a success rate superior to the total loss rate and be sufficient to earn consistently more than the opening balance month on month.
Second, we don't want to seek colossal returns. You will find that your programme will need continuous optimisation as the success rate steadily builds your account, and you will find trade sizing and diversity become ever more crucial. Compounding your trade sizing increases your underlying risk in pounds and pence, even though your percentages remain static, when you should really be looking to decrease risk as your trade size grows.
Be sure you understand what drives the marketplace you trade. Algorithmic trades do not work on hunches.
Next, you want to consider a really contentious issue for traders. Trades with high probability of success earn fewer points/pips. We might see a trade that will earn 50+ pips per day, and I have posted them on site, but they expose you to the market for longer. Most will not consider time as a risk factor, but if you trade six or seven-figure sums, you will probably reconsider that view. So would you rather trade something that gives you regularly 10pts/pips per day with incredible reliability, or something that has a lower success rate but earns, say, 50pts/pips per day while incurring far more losses or drawdowns (or both).
Consider a simple and robust algorithmic trade based
on a three-way currency pairing. Photo: iStock
The earlier posted algorithmic trade is as follows;
EURUSD rallies faster than GBPUSD rally, then EURGBP rises
EURUSD rallies slower than GBPUSD rally, then EURGBP falls
EURUSD falls and GBPUSD rallies, then EURGBP falls
EURUSD rallies and GBPUSD falls, then EURGBP rallies
EURUSD falls faster than GBPUSD falls, then EURGBP falls
EURUSD falls slower than GBPUSD fall, then EURGBP rallies
No charts are required, nor is price a concern. Your point of measure is merely how far EURUSD
moves in relation to Cable (GBPUSD
) in terms of pips or the rate of change indicator. The live trade posted for this very simple algorithm showed profits regardless of the market environs as we weren't interested in market price direction, but the correlation between each FX pairing. The beauty of running some algos is that you can at times ignore direction entirely and let the pure mathematics — or, if you like, "the force" — be with you!
This is a pure basic structure and can be enhanced easily using ratios between the currencies for trade entry criteria and exit criteria. Don't be fooled by its simplicity as it is robust. Furthermore this is not the only three-way currency pairing you can establish and benefit from.
Determine what you want to measure and how you will trade it: price/spread/ROC and likewise return on capital employed. You really need to consider what you expect to earn and then determine if the trade structure is worthwhile. It would after all be pointless to trade an illiquid market that earns less than the "risk-free return rate" you could get from, say, a savings account, which (on balance) has no risks.
Stops need setting as a matter of default, but with algorithmic scenarios you can preset the classic systems or trail in monetary values or percentage values or even ratio analysis for spread trades. Be sure you know what optimises your returns.
Eliminate lag as much as possible. Site providers will testify that the underlying use of lagging averages is a major source of curve-to-fit issues that cause many successful theoretical trade strategies to fail. By all means enter via these, if that's your preference, but they usually require a bigger stop to perform reliably.
Time-stamp requirements and globex skew are another source of trouble. When you backtest and see the colossal returns that can be made, especially with compounding, only to find that with real money in real-time situations your account is destroyed partly due to time-stamp issues. Pre-set your programme to run real-time hours only. Never try to allow a trade to run 24 hours a day, five days a week, as globex conditions rarely function as they would in normal market hours due to illiquidity and because market providers use quant models for suggested market price values throughout the closed hours of a market.
Tick backtest as normal testing could produce a curve-to-fit scenario so historic data follows the equity curve, but never earns in real time. Many traders use the mammoth markets of futures — often because they are exceedingly liquid and margin requirements are lower.
Please remember, however, that futures usually operate at a discount to the cash-market price, but sooner or later will settle at the cash-market price. Any trades opened on a futures market will incrementally day by day trade in the manner you expect, but will also adjust to the reality of a cash settlement price. The spread between futures and the cash market is often capable of destroying your trade function and also gives entirely bogus results as to success values of a strategy when you backtest. Forgetting time decay (theta) when trading futures for any period of time can be perilous.
Correlation and non-correlation structures/benefits you. Obviously we can trade highly correlated markets or equities, but when they go against you, then remember that all the highly correlated markets/equities will also go against you.
The smuggler Han Solo won his spaceship, the Millennium Falcon, in a card game.
But you shouldn't count on winning that way. Photo: iStock
The simple six-line algorithmic trade outlined above is utterly correlated, but only trades the imbalance between EURUSD and GBPUSD, which permits tighter control of your trade. Consider uncorrelated markets, such as soft commodities and the Dax. When one collapses, the other is unlikely to follow as hard or as fast, so it can slow down losses when everything goes bad in the general markets. Sometimes a defensive uncorrelated trade helps.
i could ramble on about correlation, especially when comparing equities and how fast each stock can move. But suffice it to say that, even in the highly volatile financial sector, some bank stocks move at distinctly different speeds from others, which can help to defend your account, if such protection is needed, or it can serve as a spread trade for revenue purposes.
Points of interest or points of control within markets is too big a subject to cover here (that will need to be the topic of another article), but an obvious place to start is the open range or last night's close price.
Deliberately there is no computer language here as I don't know what software/system you use, but can be posted if needed. Trade small with these functions. If the trade strategy is good, you will steadily accrue revenue.
Another subject for later is reviewing existing strategies. Ignore anyone who says the Monte Carlo technique works, as doubling your stake on each trade to ensure you make a profit is just daft. Do not be tempted by this.
Say you had seven duff trades in a row, and initially you traded $1 per point with a 50-pip stop. Then on your eighth trade, you need $128 per point with a $6,400 margin requirement, but losses incurred would be $6,350, meaning you would have to earn 50 pips to reach breakeven. Many have fallen for this scam of a strategy and it is banned in casinos or they set a maximum stake size to guarantee failure. Even George Soros tested this and found it to be useless. So, if you have more money than him, then feel free to try it.
Finally was "the force" with the algorithmic trade described above?
The bare results of this strategy running from Monday, July 11, to the time of writing, Thursday, July 14, after the Bank of England announcement, is 360 pips with no losses. All posted live.
Don't believe anyone who says the "Monte Carlo technique" works.
It doesn't. Photo: iStock
— Edited by John Acher
*fxtime is a pseudonym