Scrap Market Review — July 2026
Market analysis from TDC Ventures. Not trading advice.
The month in one line
Turkey’s scrap benchmark spent July grinding to a floor, then popping ~$5/t on freight and war-risk that mills couldn’t pass into finished steel — a cost-push, not a recovery, and we said so at the time.
Our track record this month
We publish a scored record of every forecast we make. As of 31 July, that record stands at:
- 6 of 7 resolved forecasts in range — an 86% hit rate.
- Mean absolute error: $2.81/t (median $1.76/t). Strip the single outlier below and it’s $1.24/t — inside two dollars a tonne on a benchmark that moves in fives.
- Direction accuracy: 71%.
July itself resolved four of those forecasts (three in range, one a clear miss) and added five new forward calls now awaiting the tape. The full, unedited record — ranges, actuals, verdicts, errors — is public: Forecast Track Record. Almost no scrap desk publishes one. That is the point.
Calls that landed
“Don’t chase above $380.” Our public weekly on 25 July called the week’s $5 rally exactly for what it was: “the rise came from war-risk and shipping costs, not real demand… firm up pricing to $370–375, but do not chase above $380.” What happened next: Turkey’s benchmark HMS held $374 and the US-origin transaction index $377.76, then froze for the entire following week and never tested $380. Anyone who chased the top of the range paid up for a move that stalled.
Turkey HMS, mid-July. Our forecast for the 17 July print called $362–370 (Kallanish) and $359–367 (Fastmarkets); the tape printed $368.0 and $364.76 — both in range, errors of +$2.00 and +$1.76/t. The bearish transmission we modelled (Europe’s marginal cargoes dragging the benchmark down) was real; it was simply a shade slower than we timed it.
The high-water mark. Our single most accurate call to date — Turkey shredded, resolved in early June — landed at $0.50/t off central. The mechanical shredded-to-HMS spread we built the forecast on held almost to the dollar.
Where we missed — and what it taught us
Our worst call of the record was Pakistan, 17 July. We read the CFR Port Qasim shredded index at $385–394 and predicted a drift lower. Instead it jumped +$12.33 in a single session to $402.23 — the largest one-day move in that index’s history — as Pakistani mills concluded a wave of restocking bookings. We got both the level and the direction wrong: +$12.23/t, the only double-digit miss on our board.
Two lessons, and we adopted both. The analytical one: a cost-supported floor is not just a floor — it is a launch pad. When an index has slid onto a level that freight and collection costs won’t let it break, a demand-led restocking bounce is a live scenario, not a tail risk; the upper bound of the forecast has to accommodate it. The process one: never score a forecast against a print that lands before its target date — an interim number briefly made this look like our best call ever, before the real print erased it. Both fixes are now standing rules, and the pattern they address — us forecasting Pakistan too low off a floor — showed up again, much smaller, on our 28 July call, which hit its range and under-shot by $0.75. A bias you can name is a bias you can correct.
The month’s market arc
July opened bearish. Turkey’s deep-sea scrap benchmark fell about $11/t in the first week with no floor in sight, then stopped falling and based near $368/t for the better part of two weeks — a standoff between resistant sellers and mills unwilling to bid up, not a confirmed bottom.
The back half brought the pop. Higher deep-sea freight — driven by US–Iran tension around the Strait of Hormuz and, later in the month, Houthi attacks on Red Sea shipping — let European and US suppliers declare the prior lows “unrepeatable” and lift offers. Turkey’s benchmark rose ~$5 to $373–374, and the US-origin index broke +$9.89 to $374.65. But the tell was in finished steel: over the same window Turkish hot-rolled coil ex-works fell ~$20/t. When a mill’s input rises while its output falls, it is absorbing a cost, not answering demand — and by month-end the rally had frozen for a week with no mill buying to confirm it.
Two other stories mattered. Iron ore round-tripped near $100 mid-month, then plunged $3.82 to $95.29 on 30 July as a Chinese ferrous sell-off accelerated and steel futures fell with it — a bearish macro undertone the scrap complex hadn’t yet priced. And in non-ferrous, copper tightened hard: an LME cash-to-three-month backwardation, open inventory at a six-month low, a Yangshan import premium jumping to $115/t from $59 in June, and copper breaking to $13,770/t on 31 July — a two-sided squeeze pulled by both US tariff-driven stockpiling and Chinese restocking.
Underpinning the Pakistan story: June scrap imports into Pakistan surged +24% month-on-month and +65% year-on-year — the restock wave that set the tape moving on 17 July, and the reason origin-market weakness kept failing to drag Pakistan down.
Structural signals we’re tracking
The EU’s non-OECD export regime lands May 2027. Under the revised EU Waste Shipment Regulation, recyclable-metal exports to non-OECD countries may be prohibited unless the destination country sits on an EU-approved list — eligibility becomes a question about the buyer’s country, not just the seller. India has already sought relief. For anyone moving European scrap into South Asia, this reshapes origin economics well before mid-2027; the market is not pricing it yet.
The copper squeeze is structural, not positional. A cash backwardation is transacted evidence of prompt tightness, not a bet — and it’s corroborated by six-month-low LME stocks, a doubling of the Yangshan premium, and roughly 58% of visible global copper stocks now sitting in CME warehouses on US tariff uncertainty. Copper-bearing scrap (cable, compressors, brass) carries firmer support than the ferrous complex right now.
Pakistani restocking beats origin transmission. Twice this month our Pakistan calls under-shot because we ranked Turkey’s move above Pakistan’s own restocking cycle. The correction — domestic mill re-engagement is the reliable price-setter, cost-push from Turkey the unreliable half — is now built into how we forecast that market.
What we’re watching next month
Three specific, falsifiable calls for August — the ones this review will be scored against next month:
- Turkey shredded CFR (~$394 now): does the freeze hold, break higher if freight sticks, or give back the $5 if freight normalises? We’re calling a consolidation near $390–400, with the first real mill bid as the tiebreaker.
- Pakistan shred index (~$413 after four straight gains): we expect it to plateau, not extend — the leader (Turkey) has stopped rising, so the catch-up is nearly done. A fresh restock wave is the upside risk we’re explicitly leaving room for this time.
- Iron ore sub-$96 and the copper backwardation: does the ferrous sell-off deepen and pressure the metallics complex, and does copper’s prompt squeeze hold or unwind back into contango? These two spreads will tell us more than any outright price.
How we forecast
We anchor to observed transaction levels in our own flow, cross-check against the leading indicators that reliably precede each market (Rotterdam leads Turkey by ~12 days; Turkey leads Pakistan by ~10–14), and score every call against the first print on or after its target date — publicly, misses included. Rigor you can check is the only kind worth claiming.
See the numbers behind this review: Forecast Track Record.