Saturday 3pm Acca Tips: How ArtificialAcca Picks Weekend Legs

ArtificialAcca builds Saturday 3pm accas by starting from the market, not from opinions. For each fixture it takes a de-vigged bookmaker price, adds a goals model, blends the two with the market weighted most heavily, and then searches for the best set of legs that fits your settings. This post walks through that process for a typical Saturday slate and is honest about what it does not do.

It does not predict winners with certainty. ArtificialAcca is new, so there is no long track record to quote. What we can show is the method, and the results page will show how every daily acca settles.

Why Saturday 3pm is a special slate#

A Saturday afternoon has the largest block of simultaneous matches in English football, and the 3pm kick-off is the traditional slot. In the UK, those matches are not shown live on TV because of the long-standing blackout. For an acca builder, the two things that matter are:

• Choice. Lots of fixtures means more legs to pick from and less pressure to include one you do not like.
• Simultaneous kick-offs. All your legs are settled within a couple of hours, and you cannot react to one result before the next kick-off.

Plenty of fixtures is a good thing for selection. It does not reduce the margin on each leg or change the maths of combining probabilities. If you want the numbers, read the hidden cost of accumulators.

Step 1: Pull the slate#

The ingest job runs every three hours. It fetches fixtures and odds from API-Football for today and the next two days, and for fixtures with odds in covered leagues it also fetches the team statistics that supply season goal averages. Everything is stored so the picks for the weekend are ready by Friday.

Step 2: Build a probability for every market#

For each fixture we calculate probabilities for match result, double chance, both teams to score, and over/under 1.5, 2.5 and 3.5 goals.

• Market probability. The bookmaker prices with the margin removed. If Pinnacle quotes the whole market we use its price. If not we take the median of the de-vigged prices across bookmakers. See how to remove the margin.
• Goals-model probability. A Poisson model fed with season goal averages at the relevant venue. Poisson models for football goals explains it and its limits.
• Blend. A weighted combination where the market carries at least 0.7 of the weight, with the Poisson share cut for thin samples and for strong disagreement, clamped between 2% and 98%. We do not use API-Football's own win/draw/win percentages, because they come in coarse buckets rather than real probabilities.

The market is weighted most heavily because the market contains information that the goals model does not have. A price from a sharp bookmaker reflects a lot of money and a lot of research. We treat our own model as a correction to the market, not a replacement for it. How ArtificialAcca's model works goes through the details.

Step 3: Find the edge#

For each leg we take the best available price across bookmakers and compute:

Edge = blended probability × best odds − 1

A leg with a positive edge means the model thinks the best price is higher than the true odds. An edge is only as reliable as the probability behind it. The market has been picking up the same news as us, so most edges are small, and a small edge is the likely outcome even when the model is right.

An example with made-up numbers: blended probability 46.8% on a home win at best odds of 2.25. The edge is 0.468 × 2.25 − 1 = +5.3%. If the true chance is 44%, the edge is actually −1.0%. That is the risk in every edge number we publish.

Step 4: Choose the legs#

The generator is a pure function. It follows these rules:

1. One leg per fixture.
2. At most two legs per league, unless you filter to fewer leagues.
3. Each leg's odds must be within your minimum and maximum.
4. It uses a beam search, which keeps the best partial combinations as it adds each leg.
5. The risk setting decides the target: "safe" maximises the combined probability (the sum of the logs of the leg probabilities), "balanced" does the same with a penalty for missing your target odds, and "value" maximises expected value.
6. Locked legs stay, excluded fixtures are skipped, and regenerating adds seeded variation so you do not get the same acca every time.

For a Saturday slate, the league cap does most of the work. If the early window has eight Premier League and Championship games, an acca cannot be eight English legs. It has to draw from different competitions.

What the model cannot see#

This is where honesty matters.

• Late team news. Ingest runs every three hours, and clubs name lineups about an hour before kick-off. If a key player is ruled out at 2pm, our probabilities will not have caught up, though the bookmaker's price probably will have moved. Check the price before you bet.
• Match context. Motivation, weather, a manager's tactical change and a derby's emotional temperature are not directly modelled.
• Correlated risk. The model assumes the matches are independent. A weather-hit afternoon is the type of event that violates that.
• Estimation error. Venue goal averages are crude, and early-season samples are noisy.

You should treat any pick as a starting point for your own judgement.

A sensible Saturday routine#

1. Generate the acca on Friday or Saturday morning to see the options.
2. Check team news about an hour before kick-off and look at how prices have moved.
3. Drop any leg where news has changed the picture. Regenerate or swap it.
4. Compare prices across bookmakers, because every point of price helps on every leg.
5. Choose a fixed stake in advance. See the staking guide.
6. Keep the leg count sensible. How many legs should an acca have has the table.

For league-specific thinking, look at Premier League acca tips and the Championship page. The ArtificialAcca generator lets you set the time window to match the Saturday 3pm block.

A worked example of the league cap#

Suppose Saturday's early slate has ten fixtures: four Premier League, three Championship, two Scottish Premiership and one League One. The generator takes at most one leg per fixture and at most two per league unless you filter to fewer leagues.

A six-leg acca is possible, but an eight-leg acca is not, however attractive the prices look. The cap forces diversification, and it can cost you a good leg: if three of the best-priced edges are in one league, only two can be used. We accept that trade. Three legs sharing one league and one afternoon share a lot of risk.

If you filter to a single league, the cap relaxes because you have asked for a narrower slate. Then the leg limit on odds and the one-per-fixture rule do the work.

The time window#

The generator lets you set a time window, so a Saturday 3pm acca can be limited to those kick-offs. A tight window has an advantage: every leg is settled in the same couple of hours, and you will not be waiting on a late kick-off for a result. It also has a cost, because fewer fixtures means less choice. If the window is too narrow to find legs with a sensible edge, widen it rather than accept a weak leg.

18+ only. Weekend accumulators lose far more often than they win, and our picks are probabilities, not promises. If gambling is becoming a problem, visit BeGambleAware.org or call 0808 8020 133 for free, confidential advice.