Fantacalcio Auction Tool

fishertiger

Local-first advisor for Classic Fantacalcio Serie A auctions

fishertiger builds player projections from historical Serie A data, prices the auction market, and supports a live draft with real-time advice. Everything runs locally against your own league rules — no account, no cloud dataset.

Python + ReactPlayer projectionsLive auction advisorMonte Carlo simulation

How a session flows

Generate

The pipeline turns raw Serie A sources and FVM listings into weighted player projections for the configured league.

Draft

The live auction screen tracks rosters and budgets as they're picked, and surfaces role-aware advice pick by pick.

Simulate

A 1,000-iteration Monte Carlo run projects the season for the drafted rosters using the league's own matchday rules.

Core capabilities

Everything a Classic Fantacalcio auction needs, before and during the draft

fishertiger combines projection modeling, market pricing, and live guidance so decisions during the auction are backed by data instead of gut feeling.

Weighted player projections

Historical appearances are weighted 60/30/10 from newest to oldest season, normalized to a 75-minute rated appearance and adjusted for penalty duty and European rotation.

Live auction advice

During the draft, advice values the best fixture-aware lineup and missing-vote coverage rather than simply summing every rostered player's projection.

Goalkeeper hierarchy modeling

Same-club goalkeepers form one capped coverage unit driven by explicit starter/backup hierarchy, while goalkeepers from different clubs add fallback and rotation value.

Season Monte Carlo simulation

A reproducible 1,000-iteration simulation projects full-season outcomes for drafted rosters using the configured bench, substitution rules, and league calendar.

Model

Built on documented assumptions, not black-box scores

Every projection and valuation choice fishertiger makes is written down, so advice can be checked against the rule that produced it.

  • Role budgets default to P 7% / D 18% / C 25% / A 50% of the FVM-weighted pool, editable per profile.
  • Availability priors of 85%, 55%, and 15% for TITOLARE, BALLOTTAGGIO, and RISERVA blend with historical data once it exists.
  • Only an explicit confirmed-inactive signal excludes a player; low probability or missing tiers reduce value instead of hard-excluding.
  • Fixture projections vary by opponent strength and home/away status while preserving each player's seasonal mean.

Local-first

Your league data never leaves your machine

fishertiger runs a local API and a local web client; there is no hosted backend or account system.

  • The private league calendar (calendario_lega.xlsx) is the only excluded input and is never committed to the repository.
  • Generated datasets and simulations are written to a data/processed folder scoped by profile and season on disk.
  • config/default_profile.json is the only public profile — anything league-specific stays local.

FAQ

Frequently asked questions about fishertiger

What the tool models, and what it deliberately leaves to the league itself.

What kind of Fantacalcio does it support?

Classic Fantacalcio Serie A auctions. The source calendar is always validated as 20 teams and 380 matches; the fantasy league can use a shorter configured interval within that season.

Do I need my real league's calendar to use it?

No — the included sources are enough to generate the dashboard, projections, and auction tools. A private calendario_lega.xlsx upload is only required for the season-level simulation.

Does it just sum player projections to rank rosters?

No. Advice values the best fixture-aware lineup and missing-vote coverage, including capped goalkeeper coverage units, rather than adding up every rostered player's projection.

Is it open source?

Yes, the software is MIT licensed and the structured base data in data/raw/ is CC BY 4.0.

Draft with a model, not a hunch

Clone it, generate the dataset from the included sources, and run a live auction or a full season simulation on your own machine.

Open on GitHub