Dogecoin Weekly Intelligence — September 28–October 4, 2026

Five questions for a scoring week

Dogecoin entered September 28–October 4 with five open checks and left it with three resolved — all three in the direction our criteria pre-committed to, which is the only kind of scoring that matters. The price barely moved (−1.0% to $0.09592); the series underneath it moved a lot. The transaction series broke the floor we published, the difficulty series broke the tolerance we published, and the ETF channel delivered the exact in-between week our criterion said to watch for. Below: the scores, the five questions re-answered with dated verdicts, and the sixth question the week forced on us — what happens to a proof-of-work chain when its hash leaves — which this week's deep dive takes up in full.

Last week's checks, scored

Last edition we named five checks. Here is how each resolved against the numbers we published thresholds for:

The honest headline: the two checks with teeth both fired. We wrote on September 27 that a drop under 23,000 transactions would reclassify the squeeze week as churn — it printed 17,860, and the reclassification is now the standing read. We wrote that a close below $0.095 would downgrade the $0.10 breakout to a failed attempt — six closes printed below it, and the downgrade is mechanical, not editorial. The security-budget check in our falsification watchlist also fired, at fourteen times its threshold; it gets its own section below. One check stayed unresolved, one was untestable for lack of a source, one was overtaken by the tape. That is a clean, uncomfortable week of scorekeeping.

Question 1: Was the $0.10 breakout a regime change?

The published criterion answers first: a close below $0.095 — the neckline of the double-bottom breakout and the 200-day EMA zone — 'would downgrade it to a failed attempt.' Six of the week's seven closes printed below $0.095, and the week's only close above it ($0.09592 on Sunday) still sat 4.1% below $0.10. The week's own tape supplies the counterpoint: the Sunday +3.4% session was the week's best, and the close recovered the entire mid-week dip — momentum, but without the level.

Verdict (dated Oct 4): a failed attempt, per our own pre-committed threshold. The check: a daily close above $0.0969 — the W39 pivot the last two weeks closed on — reopens the breakout read; a weekly close below $0.090, the floor this week's low (October 2, $0.09033) defended, opens the range-bottom conversation. Until one prints, the honest description is 'a range between ninety cents and ten dimes with the bias fading'.

Question 2: Is the ETF channel a demand line?

The criterion was built for exactly this week: after the record $2.89M week, we wrote, 'a repeat $0.00 week would mark this as squeeze-adjacent noise; a second consecutive record-scale week would upgrade the channel.' What printed was neither — roughly +$0.3 million net (about $879K in on September 29, $551K out on September 30, nothing after) in a week when price fell 1.0%. The monthly receipt behind it is genuinely notable: $3.71 million for September, the best month since January 2026 (SoSoValue via 24/7 Wall St.), led by Grayscale's GDOG while 21Shares' TDOG ran −$593K. And the channel's shape is about to change: Bitwise's BWOW (about $801,400 in assets) stops trading October 14 and distributes cash October 22.

Verdict (dated Oct 4): present at noise scale — alive, unproven, about to shrink. The check: flows with a flat price two weeks running would argue the channel no longer needs a rally to escort it (September's whole positive month, notably, came without a sustained rally); the post-BWOW weeks measure what a two-fund channel does. We publish whichever prints.

Question 3: Is usage recovering?

The criterion answered before we could: 17,860 transactions a day, down 25.9%, is below the 23,000 line we published as the reclassification trigger — so September 21–27's +1.9% is now formally scored as squeeze churn, and the question changes. The new geography: the series is at a dataset low (the prior low was this summer's 26,210 week), the decline from the August 17–23 peak of 31,562 now totals 43.4%, and the composition offers one thin comfort — the fall was orderly all week, with no churn-spike signature, rather than a one-day event. Active addresses fell too (34,985, −7.6%, also a dataset low). Dogecoin's first smart-contract testnet (DogeOS Chikyu, September 30) cannot move this series and has not: testnets do not touch mainnet blockspace.

Verdict (dated Oct 4): not recovering — in freefall, now floor-seeking. The new checks, published in advance: a weekly average below 17,000 extends the decline; above 21,000 starts a base; in between is floor-testing, and we will report the number, not the mood.

Question 4: Is $0.10 support, resistance, or a magnet?

The level stopped being the question this week — price left it alone entirely. The closest approach was Wednesday's intraday high ($0.0981, September 30), still 1.9% short; the close sat 4.1% below. What the week actually tested was $0.090, and it held: the October 2 intraday low of $0.09033 survived a market-wide flush (Bitcoin under $84,000, roughly $570M of liquidations on October 3) and the Sunday rebound paid the week back in full. The analysts' map from last week's coverage — $0.107–$0.118 upside, $0.095 neckline, 200-day EMA near $0.093 — has been overtaken by the middle: price spent the week living between the neckline and the floor.

Verdict (dated Oct 4): resistance, untested this week; the live conversation is $0.090. The check: the $0.0969 pivot, as above. A weekly close under $0.090 is the honest bearish trigger — it would be the first such close since mid-August — and we will not soften it if it prints.

Question 5: What are the whales actually doing?

Less than nothing is verifiable this week. The new claim — 545 million DOGE accumulated in 24 hours, ~$50.6M — comes from one X account with no named data provider and no wallet threshold, and CoinMarketCap's community analysis says as much. The prior week's 1.14 billion DOGE / $112M claim (Ali Charts via Blockonomi, named source) was still being recirculated on October 4 (Cointurk), but the exchange netflow series our check runs on printed nothing we can source. The rule holds even when it returns nothing: a claim without a tracker is a screenshot, and we do not score screenshots.

Verdict (dated Oct 4): unverifiable, unchanged. The check is unchanged too — large-address exchange netflows, two weeks running — with the added note that if no source prints it, we will retire the check as untestable rather than quietly keep it alive.

The check we didn't publish: the security budget

Our falsification watchlist carried one row about the mining layer: 'difficulty and hashrate holding ±1% while price moves ±10%', with the evidence-against branch reading 'a >2% single-week difficulty move against flat price.' The week printed the against-branch at fourteen times its threshold: day-end difficulty fell 18.2% week over week (trough −28.0% on October 2), implied hashrate fell 18.1%, and price fell 1.0%. The watchlist exists precisely so this paragraph could be written without adjectives — the check fired, on schedule, against the quiet backdrop it was designed for. What it means, mechanically, is this week's deep dive: the block clock never wobbled (1,363 blocks/day, 63.4 seconds), the subsidy never changed, and the week's final two day-end prints had already rebounded +13.7% off the trough. A security margin that can leave this fast can also return this fast — which is a property of merge-mined Scrypt, and the reason we measure the path rather than the endpoints.

Verdicts at a glance

The falsification watchlist

The standing rules, published in advance so next week can be graded rather than reinterpreted:

Where this week sits in the five-week range

Context for the verdicts, so no single week gets read in isolation. The last five weekly closes, and the two series that carry information:

Three patterns worth naming. First, the pivot held, barely: $0.09592 is 1.0% under the W39 close, and the last two weeks now read as a double top at the $0.096–0.097 area rather than a launchpad. Second, the usage series broke its structure: eight weeks oscillating between 23,660 and 31,562 ended with a print 24% below the bottom of that band — this is the series' widest single-week departure on record. Third, the divergence is the story: price held 99% of its ground while the two on-chain series fell to their lows. Markets can do that for a while; the series that closes the gap is the one to watch.

Synthesis

The through-line of the week: the checks did the work. Every major read this week came from a threshold we published in advance — the $0.095 neckline (six closes through it), the 23,000 line (breached by 22%), the ±2% difficulty tolerance (exceeded by a factor of fourteen), the flow criterion that anticipated an in-between week (it got one). Nothing required interpretation at the moment of decision; the interpretations were all paid for in advance. That is the method working under the least glamorous possible tape: a −1.0% week with no events, in which the on-chain structure moved more than the price did. Next week the calendar has teeth — the BWOW clock (October 14), a difficulty rebound needing a third print, and a usage series that must find a floor. We will run the checks and publish the score.

Appendix: data snapshot