Skip to content
#small language model Open access

The %K of Stochastics Is the Position in the Range Itself: A Touch Rate of 94.0897% Must Not Be Compared with 50% but with the Position, 90.3830% [F006]

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

Abstract

ストキャスティクスの %K は「直近の高安のどこに終値があるか」を 0 から 100 で表し、「80 以上で買われすぎ、20 以下で売られすぎ」と読まれる。この %K が位置の言い換え以上のものを持っているかを、ティックデータで測った。 まず %K と「(終値 − 直近の安値)÷(直近の高値 − 直近の安値)」を突き合わせると、四銘柄あわせて 144,463 回の比較で差の最大が 0.00e+00 で、一件も違わない。名前が違うだけの同じ式である。これは定義からの帰結であり、本稿が加えたのは件数で確かめたことだけである。 そのうえで、%K の五区分ごとに「直近14本の高値に安値より先に触れた割合」を測り、その隣に位置の平均を置いた。%K が 80 以上の区分では到達率が 94.0897% に達するが、同じ区分の位置の平均は 90.3830% であり、差は 3.7068 ポイントにすぎない。到達率 94.0897% を 50% と比べると 44 ポイントの差に見えるが、比べる相手は位置である。どちらを分母に置くかで話が十倍以上変わる。 二十の区分すべてで、到達率と位置の差の中央は +1.0445 ポイント(最小 −4.0244・最大 +6.2701)である。位置そのものが 9.6382% から 90.3830% まで動くのと比べると、残りは一桁以上小さい。差には形があり、下の二区分は四銘柄すべて負、上の二区分は四銘柄すべて正で、位置が高いほど位置が言う以上に上へ行く。 この残りは合成乱数では再現されない。足の中まで独立な正規増分で作った系列では、同じ差が 0.1001 から 0.1946 ポイントである。乱数にも同じ打ち切り(どちらの端にも触れなければ外す)をかけてあるので、差は打ち切りの偏りでは説明できない。残りは実在する。ただし位置そのものより一桁以上小さい。 方向の情報(信号の符号と14本先までの値動きの積の平均・単位 pips)は、評価期間で日次クラスタの t の絶対値が 2 以上のセルが 4 のうち 2(偶然なら およそ0.2)だが、14本ごとに一件だけ残して重複を除くと 4 のうち 0 になり、EURUSD は 3.65 から −0.32 へ符号も変わる。時間帯を重ならない五区分に分けると 20 のうち 6 が |t|≥2 になるが、この表は重複の補正を入れていないため、そのまま読めない。 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。ストキャスティクスの定義も任意停止定理もランダムウォークも既知である。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。測ったのは方向の情報であって、手数料・約定・資金管理を含む売買の成績ではない。%D と %K の交差は位置の話ではないので扱っていない。決着しなかった場面を外しており、外した件数そのものは報告していない。重複を除くと件数が 506 から 565 まで減るため、「有意でない」は「無い」ではなく「この件数では見えない」である。測定は日中・少数銘柄に限られ、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。 ----- The %K of stochastics expresses, on a scale of 0 to 100, where the close sits between the recent high and low, and is read as "80 or above is overbought, 20 or below is oversold". This paper measures, on tick data, whether %K carries anything beyond a restatement of that position. Comparing %K with "(close − recent low) ÷ (recent high − recent low)" gives a maximum difference of 0.00e+00 across 144,463 comparisons in four series: not one differs. They are the same expression under two names. This follows from the definition, and what this paper adds is only to confirm it by count. The share touching the highest high of the last 14 bars before the lowest low is then measured in five bins of %K, with the mean position placed beside it. In the bin where %K is 80 or above the touch rate reaches 94.0897%, but the mean position in that bin is 90.3830%, a gap of only 3.7068 points. Against 50% the touch rate looks like a gap of 44 points; the thing to compare it with is the position. Which denominator is chosen changes the story more than tenfold. Across all twenty bins the median gap between touch rate and position is +1.0445 points (minimum −4.0244, maximum +6.2701). Against a position that itself runs from 9.6382% to 90.3830%, the remainder is more than an order of magnitude smaller. The gap has a shape: the two lower bins are negative in all four series and the two upper bins positive in all four, so the higher the position, the further above what the position says price goes. That remainder is not reproduced by a synthetic random walk. On series built from independent normal increments down to the movement inside each bar, the same gap is 0.1001 to 0.1946 points. The same truncation (dropping cases where neither edge was touched) is applied to the random walk, so the gap cannot be explained by that bias. The remainder is real, and more than an order of magnitude smaller than the position itself. Directional information — the mean of the signal's sign times the price move over the next 14 bars, in pips — gives a daily-clustered t above 2 in absolute value in 2 of the 4 cells in the evaluation period (chance gives about 0.2), but keeping one observation every 14 bars to remove the overlap gives 0 of 4, and EURUSD moves from 3.65 to −0.32, changing sign. Splitting into five non-overlapping sessions gives 6 of 20 cells above 2, but that table carries no correction for overlap and cannot be read as it stands. Scope of this paper: no new mathematical theorem and no new law is claimed. The definition of stochastics, the optional stopping theorem and random walks are all known. No real course, author, or vendor is the subject here, and none is named. What is measured is directional information, not trading performance including commissions, fills and money management. Crosses of %D and %K are not questions about position and are not treated. Cases that did not settle are dropped, and the number dropped is not reported. Removing the overlap leaves between 506 and 565 observations, so "not significant" means "not visible at this count", not "not there". The measurement is intraday and covers few instruments, and does not contradict results documented at longer holding periods. 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

Microsoft Research Blog Oct 12, 2026

Simulating student learning for better AI tutoring

StudentSim is a research system for evaluating and improving AI tutoring tools before they are tested with learners. It models how individuals may engage with tutors, giving feedback to improve tutor performance across subjects like chess, writing, and mathematics. The post Simulating student learning for better AI tutoring appeared first on Microsoft Research.

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