Hassad AI 4HF

Hassad Workflow

How the holding turns price prediction into procurement decisions — the full monthly cycle, from portfolio demand to reviewed outcomes.

16 → 8benchmarks driving 8 companies' costs

Portfolio-wide exposure

One benchmark move cascades through several companies at once — corn alone prices Alaf's rations, Al-Waha's broiler costs and Widam's Brazilian chicken imports. Only a holding-level view captures the true position.

~90%of food imported — price-taker on world markets

Imports mean timing is the lever

Qatar imports the vast majority of its food. Portfolio companies cannot move world prices — the only controllable margin is WHEN they buy, HOW MUCH they commit, and WHICH origin they choose.

16models backtested monthly, no look-ahead

Discipline over instinct

Every forecast ships with its tested track record — walk-forward MAPE and direction accuracy on that exact commodity. Decisions are sized by evidence, and every call is reviewed against actuals after the fact.

Why predict prices at all?

The portfolio cannot move world prices — and with regulated retail prices at home, cost spikes cannot simply be passed to consumers. The only lever is when, how much, and where to buy. Prediction turns that lever from guesswork into a measured decision.

The numbers this is meant to move

Purchase price vs market

target below market (< 1.00×)

Average price paid versus the period's average market price, per commodity.

How prediction moves it: Buying ahead of forecast rises and delaying into falls beats the period average.

Tracked on Prices · Cost Models

Cost avoidance

maximise, reported per decision

Dollars saved by timing and tranche decisions versus buying on the day of need.

How prediction moves it: Direct payoff of acting on direction calls — the corn example, repeated monthly.

Tracked on Cost Models · Overview

Budget variance

target near zero

Actual input costs versus each company's quarterly cost budget.

How prediction moves it: 1–3 month forecasts feed the budget before commitments — fewer quarter-end surprises.

Tracked on Cost Models · Portfolio

Forecast hit rate

beat the coin flip, honestly measured

Share of direction calls that proved right, per commodity and horizon.

How prediction moves it: The quality gate for every other KPI — measured walk-forward, no look-ahead.

Tracked on Backtesting

Forward coverage

optimal per outlook, not maximal

Share of next quarter's requirements already priced or committed.

How prediction moves it: Rising-price forecasts justify extending coverage early; falling ones keep powder dry.

Tracked on Portfolio demand book

Shock lead time

weeks, not invoice day

Days between an alert firing and the cost movement reaching an invoice.

How prediction moves it: Alerts and scenario wizards surface exposure while there is still time to act.

Tracked on Alerts · News

Without prediction With prediction

Purchase timingBuy when stock runs low — take whatever the market charges that day.Pull purchases forward before forecast rises; delay and stay spot into forecast falls.

BudgetingInput-cost spikes land as quarterly surprises, after the money is spent.1–3 month cost visibility per company feeds budgets before commitments are made.

Shock responseDroughts, disease outbreaks and freight squeezes are discovered in the invoice.Alerts and scenario wizards flag exposed benchmarks while there is still time to act.

Portfolio coordinationEight companies buy the same commodities separately, at different times and prices.The holding sees combined exposure and can time and consolidate across companies.

AccountabilityGut-feel calls with no record — nobody can say if buying early ever paid.Every call carries its forecast, the model's tested accuracy, and a post-trade review.

What one good timing call is worth — corn for Alaf

Monthly purchase

~10,000 t

Alaf runs 350,000 t/year of feed; corn is the base energy grain.

Monthly spend

~$2.2m

At ~$220/t on the live FRED corn benchmark.

Forecast flags +6% next month

≈ $130k

Cost avoided by pulling ONE month's tranche forward before the rise.

Wrong-call downside

carry cost only

Buying early costs weeks of storage; missing a spike costs the full move. The asymmetry favours acting.

Repeat across 16 benchmarks and 12 monthly cycles: timing discipline compounds into millions — even at direction accuracy modestly above a coin flip, which is exactly what the Backtesting page measures per commodity.

Illustrative arithmetic on live benchmark price levels — not a recorded trade. Real tonnages and savings depend on each company's contracts.

Who gets what

The procurement desk

A direction call plus a conviction dial (tested accuracy) for every commodity, every month — act, tranche, or wait, with evidence instead of instinct.

Portfolio company CFOs

1–3 month input-cost visibility for budgets and pricing decisions — critical where retail prices are regulated and spikes cannot be passed on quickly.

Hassad Food (holding)

One consolidated exposure picture, food-security readiness ahead of supply shocks, and an auditable record of every buying decision against its forecast.

Phase 1 — Before — aggregate & frame

Collect demand→ Portfolio

Every company submits what it needs to buy and when.

Map exposure→ Cost Models

One map of which prices hit which companies, and how hard.

Scan conditions→ Crop Calendars

Harvest timing, weather and news set the backdrop.

Phase 2 — Decide — use the prediction

Read forecast→ Forecasts

Predicted prices for the next 1–3 months, with ranges.

Check accuracy→ Backtesting

Each model's real track record on this exact commodity.

Decide & commit→ Forecasts

Buy now or wait; lock contracts or stay spot; split volume.

Phase 3 — After — execute & learn

Execute + alerts→ Alerts

Orders go out; alerts watch every committed benchmark.

Track outcome→ Overview

Compare what happened against what was forecast.

Review & learn→ Backtesting

Keep what worked, adjust what did not, restart the cycle.

Step reference

Collect portfolio demand

Each company submits its procurement calendar — tonnages and target months. Alaf's feed grains, Baladna's powder imports, Widam's meat programme, Al-Waha's feed-driven needs, QATFA's off-season imports.

Output: consolidated demand book by commodity and month

Portfolio →AlafBaladnaWidamAl-Waha+2

Map holding-wide exposure

Roll demand up against the benchmark → product → company cost-share map. One corn move hits Alaf's rations, Al-Waha's broiler costs and Widam's Brazilian imports at once — only the holding sees the total.

Output: ranked exposure list — which benchmarks matter most this cycle

Cost Models →Prices →AlafAl-WahaWidam

Scan market conditions

Where is each crop in its cycle (harvest pressure vs weather-risk window)? What are sourcing-region conditions and headlines saying? This frames whether the forecast is entering a seasonal tailwind or headwind.

Output: context notes attached to each high-exposure benchmark

Crop Calendars →Weather & Crop →News →AlafBaladnaWidamQATFA

Read the forecast

1–3 month predicted prices with confidence bands, produced by the model that tested best on each commodity at each horizon. The reasoning trail shows exactly which signals drove the call.

Output: directional view + expected range per benchmark

Forecasts →AlafBaladnaWidam

Check how much to trust it

Backtesting shows each model's real walk-forward record on THIS commodity — MAPE, direction accuracy, band coverage. High direction accuracy → act with conviction; weak record → smaller tranches, keep optionality.

Output: conviction level per forecast (act / tranche / wait)

Backtesting →AlafBaladnaWidam

Decide timing, volume & structure

Rising forecast with strong accuracy → accelerate purchases or lock longer contracts before the move. Falling → delay, stay spot, shorten commitments. Uncertain → split volume into tranches. Consider origin switches (AU vs Black Sea barley) when spreads justify it.

Output: procurement instructions per company — when, how much, what structure

Forecasts →Supply & Demand →AlafWidamBaladna

Execute and set alerts

Company procurement teams place the orders. Threshold alerts go on every benchmark just committed — if the market breaks the assumption behind the decision, the desk knows immediately.

Output: executed orders + live alert coverage on open exposure

Alerts →AlafBaladnaWidamAl-Waha

Track outcome vs forecast

At delivery, compare actuals against the forecast that justified the decision. The dashboard's live cost-impact roll-up shows what benchmark moves are doing to each company's input costs right now.

Output: realised savings / cost-avoidance per decision

Overview →Cost Models →AlafBaladnaWidamAl-Waha+1

Review and recalibrate

Monthly: which calls paid off, which models slipped on the leaderboard, what the next cycle's budget assumptions should be. Learnings feed straight back into step 1 of the next cycle.

Output: updated model trust + next cycle's planning assumptions

Backtesting →Portfolio →AlafBaladnaWidamAl-Waha+2

When geopolitics hits: the same shock, two responses

Qatar's food supply runs through chokepoints and conflict-exposed origins. Prediction infrastructure doesn't see strikes coming — it means the desk already knows its exposure, has thresholds armed, and executes a rehearsed playbook instead of scrambling.

Strait of Hormuz — US–Iran escalation

Nearly all of Qatar's seaborne food imports pass through Hormuz. An escalation reprices freight, insurance and every imported benchmark at once.

AlafBaladnaWidamAl-Waha

Without prediction

1. Headline breaks: Desk learns from the news. No ranked exposure map exists.

2. Scramble week: Days lost asking which contracts, cargoes and companies are hit.

3. Panic decisions: Buy at spiked prices — or freeze and hope it passes.

4. Invoice shock: True cost surfaces in invoices for months. No lessons captured.

With prediction

1. Alert fires: Freight and benchmark thresholds trip. Exposure list already ranked.

2. Scenario in hours: Wizard quantifies cost impact per company from the cost-share map.

3. Playbook executes: Pre-agreed tranches, alternate origins, coverage extended calmly.

4. Tracked & learned: Actuals tracked against the scenario. Playbook sharpened for next time.

2017–21 blockade: Qatar re-routed nearly all food imports overnight — the shock that made food security a board-level mandate.

No model predicts a strike. Prediction's value here is readiness: exposure already mapped, thresholds armed, playbook rehearsed.

Black Sea — Russia–Ukraine disruption

Russia and Ukraine supply roughly a quarter of world wheat exports plus major corn, barley and urea volumes — the heart of Alaf's feed book.

AlafQATFAAl-Waha

Without prediction

1. Prices gap up: Wheat +40% in weeks. Feed contracts reprice before anyone reacts.

2. Blind repricing: Which products, which companies, how much? Weeks of spreadsheet work.

3. Late hedging: Coverage extended at the top of the spike — locking in the damage.

4. Margin erosion: Feed-to-broiler cost chain absorbs the hit; budgets blown quietly.

With prediction

1. Sentiment shifts: News feed and tone scores deteriorate before the worst of the move.

2. Calendar context: Crop calendar shows what a spring vs autumn escalation threatens.

3. Timed response: Tranches placed early; origins switched (AU barley for Black Sea).

4. Costed & reviewed: Per-company impact from cost models; every call reviewed vs forecast.

Feb–Mar 2022: wheat futures +40%, corn +20% in three weeks; urea followed. Days of lead time were worth full percentage points of margin.

Escalations are partially visible: sentiment sours and calendars frame the stakes — but the edge is measured readiness, not clairvoyance.