Skip to content
#small language model Open access

The Distance to the Level Ahead Moves the Time-Outs, Not the Win Rate — The Unconditional Win Rate Moves 9.97 Points, of Which the Time-Outs Account for 27.17 While the Win Rate Among Decided Trades Moves Only 2.46 [F028]

Sep 2026 · Zenodo (CERN European Organization for Nuclear Research)

Abstract

「利確の手前に抵抗があるから、そこまで届きにくい」という言い方がある。だが「届きにくい」には二つの行き先がある。反対の柵へ行ったか、どちらにも届かなかったか(時間切れ)である。前者なら勝率が落ち、後者なら決着しないだけである。そして柵が対称でコストが無ければ、決着した中での勝率は任意停止定理により 0.5 である。水準がそこを動かすなら定理が破れていることになる。本稿の分離子は手前の水準までの距離であり、入口・柵・期限・銘柄を固定して、距離 ÷ 柵で層に分ける。水準は確定してから使う。前後 3 本の山は 3 本あとにならないと山だと分からないので、3 本待ってから使う。 四銘柄(USDJPY・EURUSD・GBPUSD・XAUUSD)の2019年から2024年までの1時間足で、全部の足の終値から買いと売りを両方建て、利確 = 損切り = 10・20 pips、期限 24 本(一日)として測った。確定した山は足の 9.57% から 9.88% である。だが宣言した設計では何も出なかった。期限 24 本では時間切れがほとんど出ず(R=10 で 0.0002)、層の効果は R=20 で +0.48 ポイント、符号すら予言と逆である。塞がれて届かないという機序が現れる余地が無かった。そこで期限 6 本の版をあとから足した。これは後出しなので、本文でもそう明記している。 期限 6 本・R=20 では、無条件勝率は手前に水準があるとき 0.3059、遠いとき 0.4057 で、9.97 ポイント動く。六つの層で一度も逆転しない。予言①どおり、手前にあるほど低い。だがその内訳は時間切れである。時間切れは 0.3774 から 0.1058 へ 27.17 ポイント動くのに、決着した中での勝率は 0.4906 から 0.4661 へ 2.46 ポイントしか動かない。決着した中での勝率はどの層でも 0.5 の近くにあり、差は −0.0094 から −0.0339 で、その大半は曖昧な足を保守的に倒したこと(F025)で説明できる。手前の水準は「どちらが勝つか」ではなく「そもそも決着するか」を変えている。 予言③と④は外れた。M =(買 + 売)÷ 2 は層で動き、端から端で −0.6838 から −1.1076 pips である。M を動かしているのは曖昧な足の割合で、F026 のとおり対称な柵での M は −(曖昧率 × 柵) にちょうど等しく、水準が遠い層ほど値動きが大きく曖昧な足が増える。だから M の動きは方向の情報ではない。そして基準線では層の効果が出ない。足の中まで独立な正規増分で作った合成系列では、無条件勝率の端から端が +0.08 ポイント、M が +0.0815 pips しか動かない。層ごとの値の相関も −0.2592 から +0.2047 と低い。合成乱数にも山はある(幾何だから当然ある)ので、それでも層が何も分けないということは、層が拾っているのは幾何ではなく市場の性質である。ただし「水準が塞いだ」とは言えない。合成乱数に無いのは山ではなくボラティリティの群れであり、手前の山が近いのは揺れが小さい期間だからでもある。本稿が示したのは層が時間切れを動かすことまでで、その機序が「水準に阻まれた」のか「静かだった」のかは分けていない。二つを分けるには、同じ静けさの中で水準の有無だけを変える測り方が要る。 この論文の位置づけ:新しい数学定理も新しい法則も主張しない。任意停止定理、トリプルバリア法、スイング水準、ランダムウォークはすべて既知である。特定の教材やその著者、販売元は主題ではなく、一切名指していない。コストを入れていないので、本稿の数字は売買の成績ではない。期限 6 本はあとから足したもので、後出しである。機序を分離しておらず、「塞がれた」のか「静かだった」のかを分けていない。水準の定義を一種類(前後 3 本の山・直近 20 個)に固定している。曖昧な足を保守的に倒しており、その規約が決着勝率を 0.5 から押し下げる。重複を除いていないので t 値を報告しておらず、有意性は論じていない。測定は日中・少数銘柄・四銘柄六年に限られ、月次の時系列モメンタムのようにより長い保有で文書化されている結果を否定するものではない。 作成にあたって:本稿の着想と内容は、著者自身の考察に基づくものです。文章の構成整理や英訳、集計に用いたスクリプトの作成には AI(大規模言語モデル)の助力を得ました。最終的な内容の解釈や誤りがあれば、それらはすべて著者の責に帰します。お気づきの点があれば、ご教示いただければ幸いです。 ----- It is often said that "there is resistance ahead of the target, so price is less likely to get there". But "less likely to get there" has two destinations: the opposite barrier, or neither (a time-out). The first lowers the win rate; the second merely leaves the trade undecided. And with symmetric barriers and no costs, optional stopping puts the win rate among decided trades at 0.5, so if a level moved that the theorem would be broken. The separator here is the distance to the level ahead: the entries, the barriers, the deadline and the instruments are held fixed while the layers are cut on distance ÷ barrier. Levels are used only once confirmed — a swing high over three bars either side is not known to be one until three bars later, and those three bars are waited. Using one-hour bars for four series (USDJPY, EURUSD, GBPUSD, XAUUSD) from 2019 to 2024, a buy and a sell were opened at the close of every bar with take-profit = stop-loss = 10 and 20 pips and a deadline of 24 bars. Confirmed swing highs are 9.57% to 9.88% of the bars. But the declared design produced nothing: with a 24-bar deadline there are almost no time-outs (0.0002 at R=10), and the layers move the win rate by only +0.48 points at R=20, with even the sign opposite to the prediction. "Blocked and does not get there" had no room to appear. A six-bar deadline was therefore added after the fact, and the text labels it as such. At six bars and R=20 the unconditional win rate is 0.3059 with a level close ahead and 0.4057 with it far, a move of 9.97 points, monotone across all six layers: as prediction ① said, closer means lower. But that move is in the time-outs. They go from 0.3774 to 0.1058, a move of 27.17 points, while the win rate among decided trades goes only from 0.4906 to 0.4661, a move of 2.46 points. The decided win rate stays near 0.5 in every layer, with gaps of −0.0094 to −0.0339, most of which is explained by resolving ambiguous bars adverse-first (F025). A level ahead changes not "which side wins" but "whether it is decided at all". Predictions ③ and ④ failed. M = (buy + sell) / 2 does move across the layers, by −0.6838 to −1.1076 pips end to end. What moves it is the share of ambiguous bars: as F026 showed, M under symmetric barriers equals −(ambiguous share × barrier) exactly, and layers with a distant level are stretches with larger movement and more ambiguous bars, so the movement in M is not directional information. And the baseline shows no layer effect at all: on a synthetic series built from independent normal increments down to the movement inside each bar, the unconditional win rate moves by +0.08 points end to end and M by +0.0815 pips, with layer-by-layer correlations of only −0.2592 to +0.2047. The synthetic series has swing highs too (geometry guarantees that), so that the layers still separate nothing there means the layers are picking up a property of the market rather than geometry. Even so, "the level blocked it" cannot be concluded: what the synthetic series lacks is not swing highs but volatility clustering, and a nearby swing high also means a stretch with small movement. What this paper shows goes as far as: the layers move the time-outs. Separating the two mechanisms needs a measurement that varies only the presence of a level within the same level of quiet. Scope of this paper: no new mathematical theorem and no new law is claimed; optional stopping, the triple-barrier method, swing levels and random walks are all known. No real course, author, or vendor is the subject here, and none is named. No costs are included, so the figures are not trading performance. The six-bar deadline was added after the fact. The mechanism is not separated, so "blocked" and "quiet" are not distinguished. One definition of a level is fixed (a swing high over three bars either side, the most recent twenty). Ambiguous bars are resolved adverse-first, and that convention pushes the decided win rate below 0.5. The overlap is not removed, so no t value is reported and significance is not discussed. The measurement is intraday and covers four instruments over six years, and does not contradict results documented at longer holding periods, such as monthly time-series momentum. Acknowledgement: The ideas and content of this paper stem from the author's own considerations. Assistance from an AI (a large language model) was used for structuring, English translation, and writing the scripts used for the measurement. Any remaining errors or misinterpretations are solely the author's. Feedback and corrections are sincerely appreciated.

View source

Similar papers

#small language model Dataset Open access Oct 2026

Socratic guiding questions in synthetic arithmetic data: matched LoRA runs (revision v2)

Supporting data, adapters, predictions and code for the article *Low-Cost LoRA Fine-Tuning of Small Language Models for Multi-Step Arithmetic Reasoning* by Jake O'Grady, Asena Isik Gürhan, Chee Fong Ting and Effirul Ramlan (University of Galway). We generated 20,000 GSM8K-derived arithmetic problems with step-by-step s...

O'Grady, Jake, Gürhan, Asena Isik, Chee, Fong Ting et al. · 465 citations
#computer vision Open access Jun 2016

Software Development in Startup Companies: The Greenfield Startup Model

The results are packaged in the Greenfield Startup Model (GSM), which explains the priority of startups to release the product as quickly as possible, and the need to shorten time-to-market, by speeding up the development through low-precision engineering activities.

Carmine Giardino, Nicolò Paternoster, M. Unterkalmsteiner et al. · 178 citations · ⚡14
#computer vision Open access Oct 2016

Software Startups - A Research Agenda

Software startup companies develop innovative, software-intensive products within limited timeframes and with few resources, searching for sustainable and scalable business models.

M. Unterkalmsteiner, P. Abrahamsson, Xiaofeng Wang et al. · 157 citations · ⚡17
#machine learning Review Open access Oct 2016

“Failures” to be celebrated: an analysis of major pivots of software startups

This study conducts a case survey study based on the secondary data of the major pivots happened in 49 software startups, and demonstrates that customer need pivot is the most common among all pivot types.

Sohaib Shahid Bajwa, Xiaofeng Wang, Anh Nguyen-Duc et al. · 127 citations · ⚡15
#computer vision Review Open access May 2015

A survey study on major technical barriers affecting the decision to adopt cloud services

The comparison of adopter and non-adopter sample reveals three potential adoption inhibitor, security, data privacy, and portability, which underlines the importance of the technical and security perspectives for research investigating the adoption of technology.

Nattakarn Phaphoom, Xiaofeng Wang, S. Samuel et al. · 111 citations · ⚡8
#computer vision Open access Feb 2018

Lean Internal Startups for Software Product Innovation in Large Companies: Enablers and Inhibitors

This study investigates how Lean internal startup facilitates software product innovation in large companies and identifies its enablers and inhibitors, and shows the potential of the method-in-action framework to investigate the Lean startup approach in non-startup context.

Henry Edison, Nina M. Smørsgård, Xiaofeng Wang et al. · 78 citations · ⚡6

Related blog posts

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.