Funded by firms that
need you to win.

What is quoted, counted

126 firms were read plan by plan, which produced 15279 sentences transcribed from their own pages and 6925 recorded silences — questions we went looking for that their own documents do not answer. Nothing is published without the address, the date and the saved copy behind it, and a checker goes back and looks for every one of those sentences inside that copy. Across the wider register, 276 of the firms’ own sentences sit under the 152 rows, and 0 of the 73 disqualified rows still carry no quotation from the firm they accuse. That last number is the honest one to watch: it can only go down.

If the best one pays us nothing

Then it stays first, and its row carries no paid link. This is not a preference we are asking you to trust: the 11 modules that produce every score, every band and every position inside a band are compiled without access to the field that holds the commercial relationship, and a check in the build fails — stopping the deploy — if that ever stops being true. The check is run against itself on every build to prove it can still go red. A firm that pays us cannot move up, because the code that decides the order cannot see that it pays. One limit, so the claim is the right size: the register of 152 firms is listed in the order its file is written in, which no code computes and therefore no check can guard. It is a list, not a ranking; the ranking is the page that is guarded.

Partner list

Every commercial relationship this register has ever had, with the date it started and, where it has ended, the date it ended. Departures are never deleted. Beside each firm sit two dates: the day we first read it, and the day the relationship began. The second must be later than the first, or the row shows no paid link at all — a rule enforced in code rather than in policy, so you can check it by reading the two numbers.

There are none. No firm in this register pays us anything today, and no row on this site carries a paid link. This line is generated from the data, not typed: it disappears by itself the day the first relationship is written down.

What we read next, decided in advance

The firewall below protects the order of this list, not the choice of what goes in it. Every score could be honest while the register as a whole was bought, just by reading the firms with affiliate programmes first — and no code can stop that, because choosing what to look at happens before any code runs. So the queue is published before the reading and not reordered: 18 of 22 done. The order comes from measured market size across six independent sources — money paid out, web traffic, organic search, review counts, conversation volume, and presence across comparison sites, that last one dropped as a size signal because being listed mostly measures having an affiliate programme. Per-firm figures are not on these rows: two traffic tools disagree by a factor of three on the same domain in the same month, and printing that would dress a judgement as a reading.

1. 1FundingPipsCFD · read
2. 2FundedNextCFD · read
3. 3Take Profit TraderFutures · read
4. 4The5ersCFD · read
5. 5My Funded FuturesFutures · read
6. 6TradeifyFutures · read
7. 7FundedNext FuturesFutures · read
8. 8Goat Funded TraderCFD · read
9. 9Breakout PropCrypto · read
10. 10Crypto Fund TraderCFD · read
11. 11HyroTraderCrypto · read
12. 12Klein FundingCrypto · read
13. 13MubiteCrypto · read
14. 14Plutus Trade BaseCrypto · read
15. 15Funding PredictsPrediction · read
16. 16PropMarketPrediction · read
17. 17PolyFundrPrediction · read
18. 18MyFundedCapitalPrediction · read
19. 19BitfundedCrypto
20. 20PolyFundedPrediction
21. 21FundedPolyPrediction
22. 22TradeXPropCrypto

Undecided slot — Alpha Futures · Uprofit · Top One Futures. Three firms sit inside the margin of error of each other on every source that measures size. Choosing now would be choosing arbitrarily and calling it a measurement.

The method has a version

Version 6.1.0, with 12 dated entries. The scores on this site come from constants — weights, floors, thresholds, gates — and changing one of them can move a firm between bands without a word of prose changing anywhere. So the build fingerprints all 15 of them: edit a weight without writing down who it moved, and the deploy stops. Every entry says what changed in the machinery and, separately, which firm it moved. The entries below 1.0.0 are marked as reconstructed, because they are: this register kept its reasoning in code comments rather than in a log until now, and presenting a reconstruction as a contemporaneous record is the exact move the site exists to catch.

• 6.1.0 · 2026-09-01

  Five places where hiding a fact could RAISE a published number, all five caught by tools/check-monotonia.mjs and none of them found by reading a firm — they came from running the check itself across the batch of 45 read this week, and it went from 40 defects to 0. `promise()` picked a firm's headline plan by comparing raw score, which let a plan we read a THIRD of outrank one we read in full by a lucky percentage; hiding a firm-level fact like the entity's name then hurt the fully-read plan's score by the same points but a bigger share of its bigger denominator, and the incomplete plan won. It now compares score times coverage — the raw points earned, which a firm-level fact removes in equal measure from every plan at once — tied on that by coverage. `foldStaged`'s own comment already promised that a concealed middle stage "does not disprove" a tightening; the code let it vanish from the worsens comparison instead, so hiding the EASY stage of a rule that got harder erased the proof it got harder. A concealed middle stage now forces the forfeit, same as the code already said it should. `foldPayoutAlso` took the minimum of whatever a plan actually disclosed, so a plan naming two consequences of a payout — one severe, one mild — scored worse than the same plan silent about the severe one; a concealed sibling in the list now forces the worst score, not the best of what is left. `survival()` checked target and room in a fixed order, so whichever ran first returned an answer for a gap that might have belonged to the OTHER side — a plan whose floor we never read started scoring the moment its target went quiet too, because the target-check ran first and never let the room-check say the floor was ours to begin with. Both checks now run before either is allowed to answer. And `ranking.ts` reused `promise()`'s headline-plan choice for its own comparator figures — ropePrice, owedCeiling, the six the reader actually compares — which answers a different question and can move for a different reason: hiding a PLAN-specific fact such as capPerRequest changed that plan's SCORE, which could hand the pointer to a different plan whose figures happened to place the firm higher. Coverage cannot move that way — the only thing every one of these checks ever does to a fact is turn a stated 'yes' into an 'unstated', and both states stay IN the ratio, so the denominator never changes. `ranking.ts` now picks its own plan by coverage alone, blind to score, which makes it blind to this by construction rather than by chance.

  Nine promise scores moved and no band did — every one of the nine kept the same letter it had, because the fix corrects WHICH plan speaks for a firm, not what any plan is worth. All nine moved the same direction, coverage up to 1.00 and score down, because all nine had been shown off a partially-read plan that scored well by avoiding a question rather than answering it: HyroTrader 7.27 to 4.36, WeGetFunded 5.38 to 3.23, Hola Prime 3.53 to 2.83, Breakout Prop 2.31 to 1.85, Lark Funding 2.69 to 2.15, Fidelcrest 2.23 to 1.34, TopOneTrader 1.84 to 1.48, FXIFY 5.19 to 5.15, City Traders Imperium 1.69 to 1.29 (coverage 0.65 to 0.85, the one short of full). Separately, thirteen firms' RANKING page now draws its comparator figures from a different plan than before — My Funded Futures, The Concept Trading, Lux Trading Firm, The5ers, Blue Guardian, Crypto Fund Trader, DayTraders.com, EmergeProfit, Funded Futures Family, Mubite, Phoenix Trader Funding, ThinkCapital and TradeDay — every one of them a tie at equal coverage between several plans, where the old code broke the tie by score and the new one breaks it by the firm's own file order; no firm's coverage number moved. No band moved from this change either. Forty defects opened this entry; none remain — `node tools/check-monotonia.mjs --all` reads 'Ninguna firma gana por esconder' for the first time since the check existed.

• 6.0.0 · 2026-08-04

  What a payout does to the rest of the account is now scored. Twenty-six plans publish a clause that changes something the moment you ask for money — the account ends, the profit goal restarts, the consistency window or the day count resets, the position size falls — and the register had recorded every one of them with its sentence and graded none of them. Shown and not scored is the gap this whole site was built against, from the other side: evidence carrying no consequence. It joins the cash axis at weight 3, graded by what the effect costs rather than by whether one exists, which is the treatment the consistency rule beside it already gets. The worst effect a plan publishes decides the grade instead of the sum, because they arrive stacked and adding them would rank a firm with three mild resets below one that closes the account. A firm that states in writing that nothing changes earns full marks, and one whose payout page simply never mentions it scores zero and stays in the denominator. Two disqualifier codes were added for the two harshest shapes, PAYOUT_ENDS_ACCOUNT and PAYOUT_RESETS_GOAL, and deliberately assigned to nobody: the enum records that a clause has a shape, a disqualifier says the shape is predatory in this firm's case, and the distance between those is a person reading the sentence.

  Seventeen grades moved and they moved in both directions, which is the tell that a real question was added rather than a penalty. Nine cash grades fell — FTMO 3 to 2, The5ers 4 to 3, FundedNext Futures 4 to 3, Take Profit Trader, The Concept Trading and Klein Funding each by one — and five rose, because a firm that puts in writing that nothing changes now earns for it: FundingPips, Breakout Prop and Crypto Fund Trader 2 to 3, Funding Predicts 3 to 4. Topstep publishes a number here and gains enough coverage to be graded on this axis at all, 3 where it had none. Six of those changes carry through to "can you keep the money". No firm changed band. The hiding test still reports zero across all 2,897 published facts.

• 5.0.0 · 2026-08-04

  Each firm is now compared against the firms selling the same product, not against whoever shares its band. A band mixes four markets, and one column showed what that costs: a futures firm charges about $0.10 per dollar of risk where a CFD firm charges about $0.33, so on the figure that answers "does this cost what it is worth", nearly every futures firm outranked nearly every CFD firm — presented as a judgement about firms, and true only about products. The measures were never the problem: inside CFD alone the same rope price runs 0.06 to 1.00, sixteen-fold, and the climb runs 0.75 to 2.40 across the crypto firms that publish it. A POSITION survives the crossing that a value cannot, so "cheapest of the CFD firms" and "cheapest of the futures firms" order a mixed band honestly and the bands did not have to be split. Two of the six figures are answered by one market only — `owedCeiling` has five readings in futures and one in CFD, `timesYourMoney` three and one — so coverage is measured against the figures a market can be compared on at all, or no crypto firm would clear the floor for lacking a futures firm's payout ceiling. A firm's own silence is still charged in full: it takes last place and keeps its weight. The market reaches the ranking as an argument, the way disqualifiers already do, so the module still cannot open the directory and the firewall check still passes.

  The order inside both bands changed. Band C is led by PolyFundr, the strongest of the four prediction firms, where the head of that band was previously decided by a comparison spanning every market at once; The5ers, Topstep and Earn2Trade reshuffled behind it. In band B the three futures firms still lead, and now that means they place well among futures rather than that futures is cheap. No firm changed band — bands come from the payment-promise score, which this does not touch. It also closed the last three monotonicity breaks on the page: MyFundedCapital could gain places by deleting the per-payout cap from its best plan, which promoted the runner-up plan and its figures, and inside a field of four prediction firms that swap no longer buys anything. `tools/check-monotonia.mjs --orden` now reports zero and gates the build.

• 4.0.0 · 2026-08-04

  A firm can no longer improve its place inside a band by removing information from the comparison. Two faults in `fitBand`, the same fault twice: a silence used to leave the average of positions, so a figure that placed a firm badly could simply be deleted — PolyFundr sat mid-table on climb at 0.50, and deleting the phase target that figure is computed from moved it from third in its band to first. And a silence used to leave the POPULATION, so a firm that was worst on a measure could hand last place to the second-worst by walking off the field: MyFundedCapital passed a competitor on rope price without one of its own numbers changing. Now an unstated figure takes last place on its measure and stays in the count, which is what every score on this site already does with a silence. Found by `tools/check-monotonia.mjs --orden`, which re-ranks the whole register under four reader profiles for every fact a firm could delete; it reported 24 breaks across seven firms and now reports three. `ranking.ts` is also signed into this fingerprint: its band gates and measures were covered from the start and the two private functions that decide the order were not.

  The order inside both bands changed. In band C, Lucid Trading moved from fourth to first and Earn2Trade from second to fifth; Topstep and four other firms crossed the fit floor and are now placed instead of trailing the band under "not enough read", because their own silences are readings. In band B, FundedNext Futures and The Concept Trading swapped second and third, and HyroTrader is now placed at the worst possible fit rather than unplaced. No firm changed band: the bands come from the payment-promise score, which this does not touch. The direction is consistent — firms that publish these figures gained on firms that do not, which is the whole point and was not true before.

• 3.0.0 · 2026-08-04

  Two figures stopped walking past a firm's silence. `survival()` took the steepest target among the phases a firm STATED, so a plan that published 8% for one phase and nothing for the next was graded on the 8% — the unpublished phase was not measured badly, it was not measured at all. `buffer()` did the same thing one field over, walking back to a shallower stage's loss limit when the deeper one was left blank. Both now stop at the silence and score it as a silence: zero, and inside the denominator, which is the rule the rest of the register already followed. Found by a new gate, `tools/check-monotonia.mjs`, which takes every one of the 2,895 facts the 29 read firms published, rewrites it one at a time as a silence, and recomputes everything the site prints — the promise score, the six axes, the three questions and the band. Before this entry, 26 of those rewrites made a number go UP. The same run also found that `cifras.ts` was not covered by this fingerprint, which is why the file that holds both functions is now signed alongside `axes.ts`.

  Two firms lost a grade on the challenge axis and with it the "will you pass" answer: HyroTrader 3 to 2 and Klein Funding 4 to 2. Both publish an evaluation drawdown and leave the funded one blank, and both were being credited with the exam figure as though it described the account they end up in. Eleven more figures across My Funded Futures, Crypto Fund Trader, Mubite and Funding Predicts changed from a number to "they do not say" — the register's own reading of those pages, now printed instead of averaged over. No firm gained. The worst case removed was FTMO, whose challenge axis rose a full point for deleting one phase target, and the pattern reached nine firms including FundingPips and FundedNext.

• 2.0.0 · 2026-08-03

  A sixth verdict key: read · graded. Not one weight, threshold or scale moved — the fingerprint above this entry is the same one 1.0.1 sealed. What changed is who wears which badge: a firm the register read plan by plan but never hand-scored on the five axes used to fall through to "not judged", and now the verdict looks the name up in the generated index of read firms and says so. The paid-link gate learned the key the same day and treats it exactly like "not judged" — the reading exists; the sentence a paid link needs does not.

  Eight firms moved from NOT JUDGED to READ · GRADED: HyroTrader, Klein Funding, Mubite, Plutus Trade Base, Funding Predicts, PropMarket, PolyFundr and MyFundedCapital. Before this key the site published a contradiction — those eight wore "not judged" on the directory while the grid and the ranking graded them on the same screen. No tier moved, no score changed, and no firm gained a paid link; checked over all 151 rows, the only verdicts that changed are those eight, in that one direction.

• 1.0.1 · 2026-08-01

  Nothing in the method changed. What changed is how much of it the version number is nailed to: the fingerprint now covers the thirty scales inside axes.ts — the functions that turn a read fact into a number out of five — by hashing that module compiled, rather than only the constants it exports. Until today an edit to any of those thirty could move a firm while the published version stood still, and the tool said so in its own header rather than hiding it.

  No firm moved, and none could have: not one line of arithmetic was touched. The entry exists because the fingerprint changed, and a fingerprint that changes without an entry is exactly what this log is for — including when the answer is "nothing moved". Proved rather than asserted: a private weight was changed on purpose, the fingerprint moved, the change was reverted and the fingerprint came back identical.

• 1.0.0 · 2026-07-30reconstructed

  The payment-promise score stopped being compared against absolute point floors while it was normalised over what we had read. Band membership is now computed from points actually earned — score multiplied by coverage.

  Lucid Trading moved from band B to band C. It published 5.1 out of 10 with two of the three heavy questions unread; in absolute points that is 3.1. The old arithmetic meant a question we never asked raised the average, which is the one thing this register promises never to allow.

• 0.9.0 · 2026-07-30reconstructed

  A challenge alone on a figure stopped counting as having won it. Placement now returns nothing when there is no rival to place against, instead of returning first place.

  Lucid Trading led band B on a figure no rival had. Our reading of it had become its advantage — a first place against nobody, averaged in as the best possible result.

• 0.8.0 · 2026-07-30reconstructed

  The coverage floor for the fit ordering is measured against our own even weighting, not against the reader’s sliders. A reader’s profile now changes the order and never what counts as enough reading.

  With one weight pushed to its maximum and the others to zero, a firm read on that axis alone reported full coverage and led its band labelled best fit, with four of five figures unread.

• 0.7.0 · 2026-07-30reconstructed

  A seventh comparable figure was added: what requesting a payout does to the loss floor.

  It had been left out with the written reason that it "barely varies among the survivors". Measured against the data, it separates four of the seven — more than a figure that was already in. The argument was made in good faith and was wrong.

• 0.6.0 · 2026-07-30reconstructed

  A firm in tier three — the band published as "we would not take this" — became ineligible for a paid link, on top of the disqualified rows that already were.

  Moved nobody: there is no paid link anywhere in the register yet. It is stricter than the plan asked for, and it costs us money rather than costing the reader anything, which is the direction every other gate in that file already errs in.

Who publishes this

This page is written and paid for by Alvaro Rhea, an individual in Quito, Ecuador, working under the name Lynoia. There is no company behind it: no registered entity, no company number, no regulator, and no jurisdiction whose courts we have agreed to. We ask every firm on this list those four questions and we score them down for each one they leave blank, so here are ours, unflattering and in the same place a reader would look for theirs. Judge this page the way it asks you to judge them — on whether each claim carries a quotation from the firm's own document, dated, with the page saved. That standard is the only thing here that does not depend on trusting us.

Corrections

If a firm on this page believes we have published something inaccurate about it, write to us and point at the specific claim. We reply within five business days. Anything we cannot stand behind against the firm's own documentation is corrected or removed, and the change is dated on that firm's row rather than edited in quietly — a claim that disappears without a trace is as hard to trust as one that never got checked. We do not remove accurate findings, and where the finding is a third party's we say whose it is and link to it.