Event-driven arbitrage means planning campaigns backward from dated events — matches, holidays, elections, platform policy changes — that predictably move traffic volume, auction prices and moderation strictness. The method: pick the event, work backward from its date, set lead times for creative and infrastructure, ramp budget into the window, and exit before the decay.
Most buyers treat events as reminders. The final is Saturday, so the budget goes up Friday night. That's not planning — that's reacting with extra steps. By Friday night the auction is already crowded, the creatives are untested, and the only lever left is paying more for the same click.
The buyers who make events work treat them like product launches: a fixed date with a production schedule in front of it. The date doesn't negotiate, so the schedule does all the moving — every asset, check and budget decision gets placed backward from it. This article is that schedule — what actually moves during an event, how far ahead each piece needs to start, and where the expensive mistakes hide.
What actually moves during an event
Three things, and only one of them is obvious.
Traffic volume. Audiences show up for the event, not for your ad. Attention concentrates on the topic for a window — a day for a fight, weeks for a World Cup — and feeds, searches and conversation all tilt the same direction. Ads that speak to the event ride that attention instead of renting generic attention.
Auction prices. Everyone else noticed the date too. Competition for the same audiences thickens as the window approaches, and CPMs climb. The closer to the peak you buy, the more you pay for identical reach. This mechanism is the whole reason the framework exists: planning early is cheaper than reacting late.
Moderation strictness. Platforms know when the world is watching. Elections, high-salience events and announced policy dates come with tighter review and faster enforcement. A campaign structure that passed review in a quiet month can bounce during an event window — a pattern the iGaming calendar piece covers from the betting side. Planning for an event means planning for its moderation climate, not just its traffic.
One more thing moves, on the way down: attention decays faster than prices do. Keep that asymmetry in mind for the budget section.
Three classes of events, three planning horizons
Not every event gives you the same warning. In practice they fall into three classes, and each class gets planned differently:
| Class | Examples | When you find out | How you plan |
|---|---|---|---|
| Scheduled | Sports fixtures, holidays, elections, announced policy dates, platform conferences | Weeks to months ahead | Full reverse schedule — the rest of this article |
| Breaking | Sudden enforcement waves, ban clusters, regulatory actions | As it happens | No planning possible; reaction playbooks and alerts |
| Derived | An announced policy update becomes an enforcement wave; an election becomes stricter moderation | Shortly after the parent event | Plan for the consequence of scheduled events, not just the date itself |
The derived class is the one most buyers miss. The parent event is on everyone's calendar; the consequence isn't. An election date is scheduled — the moderation tightening that follows it is derived. An announced platform policy update is scheduled — the enforcement wave a few weeks later is derived. You can't put an exact date on a derived event, but you can put a window on it, and a window is enough to plan around.
This is also why a flat fixture list underperforms as a planning tool: it shows the parent events and none of the consequences.
The reverse timeline: lead times that hold up
The framework in one line: the event date is fixed, so everything else gets scheduled backward from it. These lead times are working defaults from common practice — calibrate them to your team and vertical, but don't compress them by default. A schedule that only works when nothing slips is not a schedule.
- 4–6 weeks out: pick the event and the angle. Confirm the offer with your affiliate manager — terms, caps, geo coverage, any event-specific restrictions. Study what ran the last time this event happened; annual events come with last year's evidence attached.
- 2–4 weeks out: creative production. This is the long pole and the place most timelines die; the seasonal creative pipeline breaks this stage into a production schedule of its own.
- 1–2 weeks out: infrastructure. Domains, white pages, tracking — checked fresh, because reputation data from last month is stale data. Localization pass on anything that references the event.
- 3–7 days out: test flights at small spend. Validate tracking end to end, pre-load budgets, prepare account-level fallbacks.
- Event window: monitoring and a pre-decided swap plan. For sports, the window includes the event itself, not just the build-up — live attention is part of what you bought.
- Within a week after: teardown. What ran, what it cost, what to reuse next year.
For the true tentpoles — World Cup, Olympics, a Super Bowl push — double the creative and infrastructure stages. The lead times stretch; the order doesn't change.
Creative and infrastructure: the pre-event checklist
The compressed version, for the week before a major window:
- Offer terms and caps confirmed in writing, including anything event-specific.
- Event-referencing creatives checked for trademark problems — official marks, team logos and league branding are moderation magnets. Generic event language ("the big match," "finals weekend") carries most of the attention with none of the takedowns.
- One creative set per account, prepared as separate files, so multi-account setups don't trip duplicate detection.
- White pages re-checked close to launch: blacklists, Safe Browsing, SSL — the same checks you'd run for any campaign, re-run because listings change.
- A fallback angle ready to go. Event windows come with stricter review, and the time to have a backup approved is before your primary gets rejected.
Notice what isn't on the list: anything that takes weeks. If an item here can't be finished in days, it belonged to an earlier stage, and finding it now means the reverse schedule slipped upstream.
Ramping budget without buying the peak
Budget timing is where the framework pays for itself. The mechanic: auction prices rise into the event as more buyers enter, peak around the event itself, and decay after — but attention decays faster than prices. The most expensive day is usually the one just after the peak, when CPMs are still high and intent is gone.
For sports, the peak isn't a point — it's the match itself. Live attention means the window stays open while the event runs, and budgets that treat kickoff as the finish line leave the best hours unbought. Plan the in-event spend the same way you planned the build-up: a defined share of budget, a defined exit.
A common ramp pattern, again as a starting default: enter the window at 20–30% of planned daily budget during the test-flight days, scale toward full budget as the peak approaches, and pre-commit to an exit. The exit is the part buyers skip. Decide the exit condition before the window opens — a date, a CPA threshold, or both — because in the middle of a spike everything still looks like opportunity.
The asymmetry to remember: arriving three days early costs three days of slightly higher CPMs. Arriving one day late costs the cheapest third of the window. And holding budget into the decay side costs more than both.
FAQ
How far ahead should I start planning for a major event?
Four to six weeks covers the full schedule for a marquee event: angle, production, infrastructure, testing. For smaller weekly events — regular-season fixtures, fight nights — a compressed two-to-three-week version of the same schedule is usually enough.
What about breaking events I can't plan for?
You plan the response, not the event. Keep alerts on for enforcement waves and regulatory actions in your verticals, and keep a pause-and-relaunch playbook ready. The goal is reacting in hours, not discovering the event from your own account metrics days later.
Do holidays work the same way as sports events?
The mechanics are identical — fixed date, build-up, peak, decay — but the windows are wider and the traffic is less spiky. Holiday campaigns also compete with mainstream e-commerce budgets, which pushes auction prices harder than most sports events do.
How do I know which events matter for my vertical?
Filter, don't collect. Pick the two or three event types your audience actually follows and ignore the rest of the calendar. A betting audience and a nutra audience can both work "seasonal events" while sharing almost no dates.
Put your next 30 days on a timeline
The framework above needs one input: the dates. ArbCalendar is a free event feed built for exactly this — scheduled, breaking and derived events, filtered by vertical and region, every card carrying its primary source, an impact direction (boost, throttle or neutral), a confidence label and a recommended action window.
Set a Telegram subscription with your verticals and a lead time of up to 30 days, and the alert arrives before the window opens — which is the entire difference between planning an event and being surprised by it.