Algorithmic Trading: Who Is Really Moving the Price?

[Fairman]

Algorithmic Trading: Who Is Really Moving the Price?

A large share of forex volume today comes from computer-driven strategies. Understanding what they do helps explain behaviors that seem strange to human traders.

Algorithms in forex serve several purposes. Execution algorithms break large orders into small pieces and send them over time to reduce market impact. Market-making algorithms stream quotes and adjust them in milliseconds. Arbitrage algorithms exploit tiny price differences between venues. And systematic strategies trade based on signals such as trends, carry, or statistical relationships.

The effects on price are mixed. Algorithms generally add liquidity and keep prices consistent across venues, tightening spreads. But they can also withdraw quickly in stress, which makes liquidity vanish and exaggerates moves, as seen in several flash events.

Patterns result. Price often moves in small jumps, reacts fast to data releases, and shows short-term bursts that reverse. Algorithms also tend to act at similar times and levels, such as around the daily fix or at key technical points.

For a retail trader, competing on speed with algorithms is not realistic. The advantage lies elsewhere: patience, selecting trades with clear risk and reward, and avoiding the timeframes where costs and speed dominate.

Be careful when strategies depend on microstructure details that algorithms operate on faster.

Practical step: examine a one-minute chart during a major news release, and note how many times price whipsawed before settling.

[Fairman]

The one minute chart during NFP is a great lesson. Watching price whip back and forth in the first seconds shows exactly why manual traders shouldn't try to compete there. The release lands at 8:30am New York, which is 1:30pm Lagos while the US is on daylight time. By the time you've read the headline number, algorithms have already reacted, reversed, and reacted again. I'd use a demo account or just watch one release with no position open and count how many times price crosses its starting level in the first minute. Most people find that convinces them to wait at least fifteen minutes before looking for anything.

[Fairman]

The flip side is useful: algos thin out quickly in stress, so the moves you see in those moments can overshoot. A lot of good entries come after that panic, not during it. What I look for is the spike to run into a clear level, like the previous day's high or low, and then stall with the spread coming back to normal. If price starts printing candles back inside the range on the five minute chart, that's the market finding its feet again, and the overshoot gives you a reasonable stop beyond the extreme. Not every spike reverses, so the stop matters. But entering after the chaos usually beats chasing during it.

[Shadow Trader]

Fairman's idea of waiting for the spike to stall at a level is the most usable part of this thread. The spread is what makes it work, because it shows when the market making algos are back.

img_13166_r1.png (85.73 KiB) Viewed 46 times

In the made up chart, EURUSD sits at 1.0852 before the release and spikes to 1.0893 inside two 5 minute bars, 13 pips through the prior day high of 1.0880. The spread widens from 0.8 to 4.5 pips at the print and is back near 1.0 within about ten minutes.

The first 5 minute close back under 1.0880 comes at 1.0874, and a stop 3 pips beyond the spike high puts it at 1.0896. That is 22 pips of risk to get back to the 1.0852 starting level, so only about 1R even when it works.

That is the hidden cost of fading an overshoot. The stop sits beyond the extreme, and the extreme is wherever thin liquidity took it, so the panic sets the risk rather than the setup.

One way round it is to wait for a lower high under the level, which tightens the stop but misses the cleanest reversals. Does anyone here size these trades smaller on purpose because the stop is so wide?

[FXS Prop Desk]

One behaviour explained by execution algorithms is the steady, one-directional movement often seen around the London and New York fixes, when large orders are being worked over a set period. Recognising that pattern helps traders avoid fading moves that are driven by large, patient flows rather than by short-term sentiment.

Algorithms also explain why quotes can vanish in milliseconds around news, as market makers pull liquidity until the new price level is clear, which retail traders experience as slippage or gaps.

[Shadow Trader]

FXS Prop Desk's point about patient flow is the other half of the spike fade discussed above. The two can look alike on a 5 minute view, but the spread usually tells them apart.

A panic spike comes with a spread that blows out and then recovers. A worked order tends to move price while the spread stays normal, because the algo is slicing into liquidity that is already there rather than finding it missing.

Picture a made up grind where EURUSD rises 16 pips over 40 minutes before the London fix, in steps of about 2 pips every 5 minutes, with no candle bigger than 4 pips and the spread sitting at 0.8 the whole time. Nothing about that looks overextended, yet fading it means standing in front of an order whose size is unknown.

The clue is often the rhythm. Execution algos that slice on a timer leave evenly spaced pushes, and when the pushes stop at a round time like 4pm London, the move often goes flat within a few bars.

So the safer fade waits for the flow to end rather than for price to look stretched. FXS Prop Desk, do you find the grind usually stops dead at the fix time, or does it bleed on for a while after?

[Shadow Trader]

Every three years, the Bank for International Settlements surveys central banks and dealers about foreign exchange activity. The results show average daily turnover, which has run into trillions of dollars, and break it down by currency pair, instrument, venue and counterparty.

Several findings surprise newcomers. Most turnover is in FX swaps, not spot. The dollar sits on one side of the vast majority of trades. London is the largest trading centre, followed by New York, with Asian hubs well behind but growing. And the retail segment, while visible online, is a small slice of the total.

For scalpers, the survey provides perspective. Retail traders operate in a market dominated by institutions, with different time horizons and goals. Understanding who the main players are, banks, hedge funds, asset managers, corporates and central banks, helps interpret price behaviour at different times of day.

Reports are free on the BIS website, and the summary charts alone are worth an evening's reading.

A point worth remembering is the effect of valuation changes. COFER shares are measured in current dollar terms, so when the dollar weakens, the euro's share rises mechanically, without anyone buying euros. Careful readers separate valuation effects from genuine reallocation.

Which finding from the survey changed how the market looks, and has it influenced trading choices?