Is pairs trading still viable for an individual?

On liquid large caps, the honest answer is that you are competing with better-capitalised participants who pay a fraction of your costs. The versions with a plausible remaining edge are the ones institutions find awkward — futures calendar spreads, less liquid pairs, and relationships that require economic understanding rather than screening.

How do I know a spread has broken rather than just stretched?

In real time you often cannot, which is exactly why the stop exists. The z-score stop and the time stop are both admissions that the distinction is unavailable at the moment it matters — they bound the cost of being wrong instead of trying to be right.

Should I add as the spread widens?

It is where the strategy's logic points and it is the single most dangerous action available. If you scale in, decide the maximum total position before entering the first tranche, and keep the z-score stop where it was — otherwise scaling and stopping are in direct contradiction.

Pairs Trading (Cointegration)Mean reversion7 min read

The break: when the spread widens and never comes back

Three failure modes: the relationship ends, the relationship was never there, and everyone else holds the same position you do.

Key takeaway

A structural break — a merger, a scandal, a divergence in the businesses — is the characteristic failure and it loses on both legs at once
Data-mined pairs pass a test by chance and have no mechanism behind them, which is why an economic rationale matters
The 2007 quant quake demonstrated that market-neutral does not mean crowding-neutral

The structural break

Two companies were tied together because they did the same thing. One is acquired at a premium, or announces an accounting restatement, or wins a contract that changes its trajectory. The relationship that made the pair a pair has ended.

The spread widens and does not come back — and because you are long one and short the other, you lose on both legs simultaneously. The structure that hedged market risk provides no protection here at all.

The z-score makes this worse before it is recognised. As the spread widens, the score rises, and a naive implementation reads that as a stronger signal and adds. That is the mechanism by which a broken pair becomes a large loss rather than a small one.

Adding as the spread widens is the strategy's logic applied to a case where the logic no longer holds. Only a hard stop distinguishes the two.

The pair that was never cointegrated

Test enough pairs and some pass at any significance level by chance. A pair discovered by screening thousands of combinations, with no economic reason behind it, has a substantial probability of being one of those.

Such a pair looks identical to a real one — same test statistic, same tidy historical spread — and behaves completely differently going forward, because there was never a mechanism pulling the two together.

August 2007, and what market-neutral does not protect against

In August 2007, quantitative market-neutral funds suffered severe losses over a few days without any obvious market event. The mechanism is now well documented: many funds held similar positions, one large participant began liquidating, the forced unwinding moved the very spreads everyone was positioned in, and the losses triggered further liquidation.

Market-neutral did not help, because the market was not the problem — the crowding was.
The positions were correlated with each other, not with the market, and nothing in a hedge ratio detects that.
Spreads that had reliably converged for years diverged simultaneously, and then largely converged again after the liquidation ended.

Anyone running a strategy that a large amount of professional capital also runs should assume their positions are correlated with everyone else's, and that this correlation appears only under stress.

The operational failures

Borrow recall. The lender wants the shares back, you close the short at market, and the remaining long leg is now an unhedged directional position.
Borrow cost spikes. A stock becoming hard to borrow can cost multiple percent annualised, which on a thin spread makes an otherwise-correct trade unprofitable.
Corporate actions. Splits, spin-offs and special dividends all distort the spread. A system that does not handle them explicitly will generate signals from the accounting.
Legging risk. Getting one leg filled and not the other, in a fast market, at exactly the moment the spread is moving.

Common questions

Is pairs trading still viable for an individual?: On liquid large caps, the honest answer is that you are competing with better-capitalised participants who pay a fraction of your costs. The versions with a plausible remaining edge are the ones institutions find awkward — futures calendar spreads, less liquid pairs, and relationships that require economic understanding rather than screening.
How do I know a spread has broken rather than just stretched?: In real time you often cannot, which is exactly why the stop exists. The z-score stop and the time stop are both admissions that the distinction is unavailable at the moment it matters — they bound the cost of being wrong instead of trying to be right.
Should I add as the spread widens?: It is where the strategy's logic points and it is the single most dangerous action available. If you scale in, decide the maximum total position before entering the first tranche, and keep the z-score stop where it was — otherwise scaling and stopping are in direct contradiction.

These are documented methods described for study. Nothing here is investment advice, a recommendation, or a claim about future returns — every system on this page has losing periods, and the pages say where.

Reading about a system is not having one.

Plutux is where you write your own rules down, test them against real data, and keep the record your memory would otherwise rewrite. Join the waitlist for early access.